Skip to main content

Full text of "Mullard Reference Manual of Transistor Circuits"

See other formats

i Mullard 





Made and printed in Great Britain by WIGHTMAN AND 
COMPANY LIMITED, 1-3 Brixton Road, London, S.W.9 



First edition 1960 

Milliard is the trade mark of Mullard Ltd. and is registered in most 
of the principal countries in the world. The issue of the informa- 
tion contained in this publication does not imply any authority 
or licence for the utilisation of any patented feature. While 
reasonable care has been taken in the preparation of this publica- 
tion, no responsibility is assumed for any consequences of its use. 

© Mullard Ltd. 1960 

Mullard Ltd., Mullard House, Torrington Place, London, W.C.1 


This manual has been produced with the active co-operation of the heads 
of departments and members of staff of the Mullard Research Laboratories, 
the Semiconductor Measurement and Application Laboratory, the Technical 
Department, and the Applications Research Laboratory. The following list 
includes those who have been of most service in the preparation of the 
manual; if the list were made more complete, it would defeat its object. 

L. G. Cripps 
L. B. Johnson 
L. H. Light 
E. Wolfendale 

E. G. Cooper 
M. J. Gay 
K. Holford 
L. E. Jannson 

H. Kemhadjian 
A. F. Newell 
J. R. Nowicki 
J. F. Pawling 
A. J. Rees 
W. L. Stephenson 
P. Tharma 
T. B. Watkins 
P. G. Williams 


The fact that the first pages of this manual were printed last allows attention to 
be drawn to the following error. On p. 92, the line before the heading which 
reads Relative Importance of f\ and f« should end with fa and not with fa . 


1. The Junction Transistor 

2. New Techniques 

3. Static Characteristic Curves 

4. Small-signal Characteristics 

5. Basic A.C. Circuits 

6. Transistor as a Current Network 

7. Bias and Stabilisation 

8. Equivalent Circuits 

9. Limiting Values 

10. Semiconductor Diodes 

11. The Phototransistor 

12. Audio Amplifier Stages 

1 3. Class A Output Stages 

14. Class B Push-pull Output Stages 

15. Hearing Aids 

16. Low-power Audio Amplifiers 

17. Public-address Amplifiers. . 

18. I.F. Amplifier for 470kc/s .. 

19. Radio Receivers 

20. 4W 500kc/s Transmitter . . 





















21. Tuned Oscillators 

22. Phase-shift Oscillators 

23. Transistor as a Switch 

24. Pulse Circuits 

25. D.C. Amplifiers 

26. D.C./D.C. Converters 

27. Protection Circuit for Stabilised D.C. Power Supply 


28. lOOOc/s Oscillator and Tuned Amplifier for H.F. Measurements 303 



Ratio detector using 2-OA79 

OA70 as video detector at 30Mc/s 

OA70 as sound detector in transistor receiver . . 

OA81 as television noise limiter 

Four OA10 in low-voltage bridge rectifier 

Basic circuit for OCP71 for unmodulated light . . 
Temperature-compensated circuit for OCP71 with d.c 
Basic circuit for OCP71 for modulated light 


Typical RC-coupled stage for OC71 

Basic OC70 preamplifier for 250V supply 

Rearrangement of 250V preamplifier for feeding valve amplifier 

60mW output stage using OC72 operating on half-supply-voltage principle 
OC26 output stage with EF98 valve driver 
OC26 output stage with OC71 driver 

Common-emitter 7W push-pull amplifier for hybrid car-radio receiver 
Common-emitter 500mW push-pull amplifier 
Split-load 540mW push-pull amplifier 
Single-ended 500mW push-pull amplifier 
Common-emitter 1W push-pull amplifier 

Four-transistor RC-coupled hearing aid . . 
Three-transistor RC-coupled hearing aid 
Three-transistor directly coupled hearing aid 

200mW audio amplifier for supply voltage of 6V 

200mW audio amplifier for supply voltage of 4-5V 

1W audio amplifier 

540mW audio amplifier 

5W high-quality audio amplifier 

15W public-address audio amplifier for supply voltage of 14V 
15W public-address audio amplifier for supply voltage of 28V 

I.F. amplifier for 470kc/s 

Portable radio receiver (original version) 
Radio receiver with improved a.g.c. 
Miniaturised version of radio receiver 











4W transmitter for 500kc/s 

10kc/s oscillator for output of 20mW 

50kc/s oscillator for output of 90mW 

42kc/s oscillator for feeding playback/record head of tape recorder 

800c/s oscillator using three-section ladder network 
Two-transistor Wien-network oscillator for about 3-4kc/s 
Three-transistor Wien-network oscillator for 15c/s to 20kc/s 



Bistable circuit 

Asymmetrical bistable circuit 

Bistable circuit using p-n-p and n-p-n combination 

Triggered blocking oscillator 

Free-running blocking oscillator (frequency divider) 

Self-gating Miller circuit 

Directly coupled d.c. amplifier for germanium transistors 
Directly coupled d.c. amplifier for silicon transistors 
Long-tailed-pair d.c. amplifier for germanium transistors 
Long-tailed-pair d.c. amplifier for silicon transistors 
Simple temperature-control circuit 
Chopper-type d.c. amplifier with silicon-diode bridge 
Single-transistor chopper and d.c. amplifier 
Balanced transistor chopper and d.c. amplifier . . 
Mechanical chopper and transformer-coupled amplifier 

Two-transformer d.c./d.c. converter 
Voltage-doubler output for d.c./d.c. converter . . 
Further modification of d.c./d.c. converter 

One-amp stabilised power supply 

Three-amp stabilised power supply including protection circuit 

1kc/s oscillator 
1kc/s tuned amplifier 










reading. Nor is it possible, in just over three-hundred pages 
of legible type, to transform the novice into a fully fledged 
circuit designer. What can be done, however, is to present the most 
important information to the largest number of users, in a way that 
can be absorbed as readily as possible. The result is the present 
Reference Manual of Transistor Circuits. 

Not only is this manual not intended for the senior designer : he will 
probably prefer not to consult it — at least when anyone else is looking. 
The aim is therefore to help all the other people who have some 
professional (or spare-time) interest in radio and electronics to realise 
the possibilities of the transistor; and to show them how to exploit 
these possibilities intelligently. The approach is not a new one, though 
it is rarely carried out on the present scale. The most useful circuit 
diagrams are chosen and revised where necessary; as much practical 
information is given as possible; and the principles of operation are 
described in such a way that the user can make at least any minor 
modifications which he may require. 

The method is not without its drawbacks. It can leave a false im- 
pression of the state of development and availability of the latest 
devices. There is always a time-lag between the introduction of a new 
device with its technical data, which enables the circuit designer to go 
ahead ; and the production of a circuit for some particular application, 
which can be put in the hands of the general user. Broadly speaking 
the circuits have been restricted to devices which are available now. 
However, the breakthrough, which the alloy-diffusion process shows 


every sign of being, means that all users must be acquainted with this 
technique, even if circuits cannot be included yet. 

Other publications exist for reporting the latest developments. In 
particular, those who subscribe to the Mullard Technical Handbook 
Service are automatically supplied with the latest technical data sheets 
as these become available. 

An important subsidiary aim has been to increase the ability of the 
average reader to appreciate technical data. The first nine chapters 
provide a background, with the main emphasis on small signals and 
audio frequencies. Subsequent chapters are more practical, though the 
necessary bits of theory are introduced where appropriate to extend 
the treatment of large signals and high frequencies. 

Short chapters on semiconductor diodes and on the OCP71 photo- 
transistor have been included because of the close relationship that 
exists between the various semiconductor devices. A few topics — such 
as some switching applications and electronic computers — have been 
judged too specialised for inclusion in a book having a wide appeal. 

The manual will be suitable for junior designers, graduate entrants 
into electronics, students of electrical engineering and physics, science 
teachers, radio service engineers, laboratory and works technicians, 
and keen hobbyists. The reader must use his judgement in choosing 
those parts which are most suitable to him. The apprentice service 
engineer may need to memorise the electrode connections, and the 
graduate entrant be more interested in equivalent circuits. 

But whether on the bench or the bookshelf, this manual is primarily 
meant as a reference source of circuits. 




Most junction transistors are, essentially, a sandwich of three layers of 
crystalline semiconductor. The manufacturing techniques and theory 
of operation are not so elementary as this simple picture would suggest, 
but a detailed knowledge of these topics is the concern of specialists, 
rather than of those who wish simply to make up transistor circuits. 

Fortunately, the operation of the transistor as a circuit element can 
be understood from a few basic ideas about its construction and 
interior mechanism. The present chapter will provide this necessary 
basis, but it must be realised that the approach has been very much 


In the normal amplifying valve, the bulb is pumped out as thoroughly 
as possible, and a getter is provided to adsorb most of the residual 
gases. The flow of electrons from cathode to anode therefore takes 
place through an almost ideal vacuum. One or more grids placed 
between anode and cathode exert a controlling influence on the 
electron stream. 

The transistor is, in essence, a crystal. It differs radically from the 
radio valve; in the valve the current flows through a near vacuum, 
necessitating an evacuated envelope and a more-or-less resilient 

The current through the transistor, on the contrary, flows through 
a solid. The envelope does not have to be evacuated, and the device 
is mechanically stronger and non-microphonic. 

A ready supply of current carriers is available at room temperature. 
The transistor does not require an auxiliary battery corresponding to 
the heater or filament supply of a valve. The transistor is therefore 
free from hum and less subject to deterioration and failure. 


In transistors and other semiconductor devices, current is carried by 
positive carriers as well as by negative electrons. 

Page 1 


The positive carriers are referred to as holes. The hole has a positive 
charge exactly equal to the negative charge of the electron. 

An electron and a hole, being of opposite sign, will attract each 
other, and the electron may eventually fill the hole. Neither can then 
take any further part in carrying current. 

This process of recombination is fundamental to the operation of 
the transistor. 


Holes and electrons normally co-exist, but in varying proportions 
depending on the impurities present. In order to control the impurity 
content and hence the conductivity with the required precision, the 
semiconductor material is first very carefully purified. Then, by the 
addition of a small amount of selected additive, holes can be made 
much more numerous than electrons, or conversely. 

Material in which holes predominate (Fig. 1) is known as p-type, 
where p stands for positive, the holes being positively charged. Where 

Fig. 1 — Semiconductor p- and n-type material 

the majority carriers are electrons, the material is n-type, n standing 
for negative. 

If holes are the majority carriers, electrons will also be present as 
minority carriers, and conversely. 


Most transistors are in the three-electrode junction construction. 
Three layers of semiconductor merge into one another to form a 

Emitter k M Collector 

Fig. 2 — Construction of three-layer (or triode) transistor 

sandwich (Fig. 2). An ohmic, that is to say non-rectifying, contact is 
made to each of the three layers. 
The central layer of the sandwich is very thin and is known as the 

Page 2 


base, and the two outer layers are the emitter and collector. The junc- 
tions between (a) base and emitter and (b) base and collector are 
known as the emitter and collector junctions. 

The transistors can be p-n-p or n-p-n, according to the additives used 
for emitter, base and collector (Fig. 3). The majority of transistors 

Fig. 3 — Sandwich arrangement of p-n-p and n-p-n layers 

available at the moment are p-n-p. It will be simpler, however, to 
consider the n-p-n type first. 

N-P-N Transistor 

The emitter, as its name suggests, corresponds roughly to the cathode 
of the electronic valve. In the n-p-n transistor, the majority carriers 
in the n-type emitter are electrons. The emitter acts as a source of 
electrons, which flow into the base when the base is biased positively 
with respect to the emitter (Fig. 4). 

The collector corresponds roughly to the anode of the valve. The 
collector of the n-p-n transistor, provided it is biased positively with 
respect to the base, absorbs electrons from the base. 


Fig. 4 — Representation of n-p-n transistor. Conventional symbol on right 

shows polarities and typical voltages. Arrow on emitter points in conventional 

direction + to — . 

Because the base is p-type, electrons only exist in the base as minority 
carriers. The electrons arriving at the collector are derived almost 
entirely from the emitter by diffusion through the base. 

Not all the electrons flowing from the emitter into the base will be 
removed from the base by the collector, as a small proportion will 
recombine with holes in the p-type base layer. 

This loss of charge in the base layer is made good by a flow of base 
current. Varying the base current varies the voltage across the emitter 
junction, and so controls the collector-emitter current. 

Page 3 


P-N-P Transistor 

The functions of emitter, base and collector are similar to those just 
described for the n-p-n transistor. This time, however, the emitter is 
p-type, and its majority carriers are holes. Holes flow from the emitter 
into the base when the base is biased negatively with respect to the 

I collector -5V 



Fig. 5 — Representation of p-n-p transistor 

emitter (Fig. 5). The holes from the emitter diffuse through the base 
and are accepted by the collector. The collector has to be biased nega- 
tively with respect to the base, in order to absorb the positive holes. 
A small proportion of the holes leaving the emitter recombine with 
electrons in the n-type base. 

Emitter and Collector Junctions 

The current carriers diffuse sideways through the base as well as 
forwards. The collector junction is therefore generally made con- 
siderably larger than the emitter junction (Fig. 2), to prevent excessive 
recombination of holes and electrons in the base. 

Some types of transistor are sufficiently symmetrical to justify then- 
use bi-directionally, that is, with what are normally the emitter and 
collector interchanged. 


The first process in transistor manufacture is the refining of germanium 
or silicon to a degree of purity beyond that attainable by normal 

Re-solidified !Molten| Solid 


Movement of molten 

-♦-Movement of impurities 

Fig. 6 — Zone refining 

chemical methods. The remaining impurity must not exceed about 
one part in ten thousand million. 

The process is called zone refining (Fig. 6). An ingot of the metal is 
drawn slowly through a tube, which is surrounded at intervals by r.f. 
heating coils. Only a few zones of the ingot are molten at any one 



time. The impurities become concentrated in the molten zones rather 
than in the solid portions. By drawing the ingot several times through 
the tube, the impurities are swept to one end, which is discarded. 

The purified metal is then grown as a single crystal. This operation 
(Fig. 7) is performed in an atmosphere of nitrogen and hydrogen to 
prevent oxidation. The metal is kept just molten by an r.f. heating 
coil, the selected additive having been added in the proportion of 

Vertical drive 
Rojtary ( 

drive \^ 




I Thermocouple 

Fig. 7 — Apparatus for growing single crystals 

about one part in a hundred million. A seed crystal, held in a rotating 
chuck, is lowered into the molten metal and then slowly raised. A single 
crystal is withdrawn from the melt with the same crystal structure as 
the seed. 

There are a number of ways of proceeding with the manufacture, but 
as most transistors are made by the alloy-junction method, only this 
method will be described in this chapter. 

To make a transistor such as the OC71, a grown germanium crystal is 
turned into slices about 4x2x0- 12mm. Each of these slices forms the 
base region of a transistor. Since the OC71 is a p-n-p transistor, the 
additive added to the melt before growing the crystal is n-type. 

A pellet of a p-type additive is placed on each side of the slice, the 
one which will form the collector being about three times the size of 
the one used for the emitter. The assembly is heated in a hydrogen 
atmosphere, until the pellets melt and dissolve some of the germanium 
from the slice. On being cooled, the pellets with the dissolved germanium 
start to solidify, and a crystal of p-type germanium grows at the solid- 
liquid interface. At a lower temperature, the rest of the pellet solidifies. 

Fig. 8 shows a cross-section of the resulting three-layer structure. 
Leads for the emitter and collector are soldered to the surplus material 

Page 5 


in the pellets, thus making non-rectifying contacts. A nickel tab is 
soldered to the slice to make the connection to the base. 

The assembly is etched to remove surface contamination. 
Finally, the assembly is covered in moistureproof grease, and her- 
metically sealed into a small glass envelope, with the leads passing 

Fig. 8 — Cross-section of p-n-p transistor 

through the glass foot. The sealing process is performed so as to avoid 
overheating the transistor. A coating of opaque paint is applied to the 
outside of the glass bulb to exclude light. 


Transistors depend for their operation upon the flow of current 
carriers (positive holes and negative electrons) within a semiconductor, 
that is, within a solid crystalline material. The presence of the two 
types of current carrier in differing concentrations, as majority and 
minority carriers, determines whether the material is p-type (excess 
holes) or n-type (excess electrons). 

Junction transistors are commonly made in the form of a sandwich 
of three layers of p-n-p, or less frequently n-p-n, material. The central 
layer, or base, is always very thin. The outer layers form the emitter 
and collector, and generally speaking are not interchangeable. The 
collector circuit is connected to the negative side of the supply for a 
p-n-p transistor, and to the positive side for an n-p-n transistor. The 
emitter circuit is connected with the opposite polarity. 

The characteristic part of the- transistor is the junction, which is a 
region of transition between p- and n-type material. A feature of the 
junction is that a current will flow readily across it at room temperature, 
when a voltage of suitable polarity is applied; preliminary heating is 
not required. 

Page 6 




Particular attention should be given to the circuit diagram. Connecting the transistor 
with the incorrect polarity may change its characteristics permanently for the 
worse, and could lead to its destruction. 


The electrode connections are given in the technical data in the form of diagrams 
such as those shown in Fig. 9. The series of diagrams in the left-hand half of the 
figure, with the appropriate shape and dimensions, apply to most small-signal and 
low-power transistors, and to some small-signal high-frequency transistors. 

The central lead is for the base, and there is a greater separation between the 
base and collector leads than between the base and emitter leads. The collector is 
further distinguished from the emitter by a spot on the adjacent part of the body. 

The diagrams in the right-hand side of Fig. 9 are given as an example of a high- 
power output transistor. Two stiff leads project through the mounting base, and these 
can be identified as the base and emitter connections by the letters B and E stamped on 


Fig. 9 — Examples of electrode connections of transistors (not to scale) 

the underside of the mounting base. The collector is electrically connected to 
the mounting base; and the collector connection is made to a tag which is held in 
contact with the top of the mounting base by one of the mounting nuts. The 
mounting base can be insulated from the chassis or heat sink as described in 
Chapter 9 (pp. 95 to 99). 


Disconnect the supply while installing transistors and, if possible, while making 
circuit adjustments. An accidental short circuit from the base to the collector supply 
line may cause sufficient current to flow to damage the transistor. Also, in some 
circuits, the supply voltage is higher than the rating of the transistor, the excess 
voltage being dropped by series resistance in the collector and/or emitter circuits. 
Short-circuiting this resistance may result in an excessive voltage being applied to 
the transistor. 

Associated components and the devices themselves should not be inserted or 

Page 7 


replaced with the power supplies connected, because of the surges which can occur, 
for instance, from the discharge of capacitors through the devices. 

Reasonable care is called for, bearing in mind the proximity of flexible leads. 


The leads are tinned to facilitate soldering. Flexible leads should not be bent nearer 
than l-5mm to the seal, otherwise the seal may be affected and the transistor no 
longer be moistureproof. 

Soldering should be completed reasonably quickly. On no account should the 
transistor be allowed to heat up during the soldering operation. The leads may be 
held in a cool pair of long-jawed pliers. The pliers act as a thermal shunt which 
virtually prevents heat from being conducted to the transistor. Where the pliers 
would require a 'third hand', the transistor may be placed inside a crocodile clip, 
the jaws of which have been filed flat to grip the leads. 

The electrical insulation between the heating element and the bit of some electrical 
soldering irons is sufficiently poor to cause a dangerously high current to flow 
through the transistor. Such irons should be disconnected during soldering. The 
plug may be arranged so that the iron is disconnected when it is picked up. 

Transistors may be dip soldered, usually at a solder temperature of 240°C for 
a maximum of 10 seconds up to a point 2mm from the seal, but otherwise according 
to the specification. 


Transistors may be mounted in any position. The only restriction on the location 
of the transistor may arise from the need to provide adequate cooling and 


Germanium and silicon are inherently sensitive to light. This light sensitivity is 
turned to useful account in the OCP71 phototransistor. 

The case of the ordinary transistor has to exclude light. For transistors in a glass 
construction — for example, the OC71 — the envelope is coated with an opaque 
paint. The paint resists normal handling, but should not be damaged. Transistors 
in a metal construction require no lightproofing. 

High-energy radiations (for example, X rays, y rays and neutrons) affect the 
behaviour of a junction, usually adversely and permanently. 


Maximum and minimum limits are set to the temperatures at which transistors 
may be stored. For most transistors, the permissible range is from —55 to +75°C. 
But the maximum storage temperature is lower for some types. 

The minimum storage temperature is only likely to be encountered in specialised 
scientific, industrial and military applications. 

Apart from the possibility of permanent mechanical damage from storage at 
temperatures below the minimum, the performance may be adversely affected by 
such low temperatures. 

Page 8 



The OC71 process, which has served as a starting point for new con- 
structions and techniques, is by no means typical of the latest designs. 
The new techniques have made possible radical extensions in perform- 
ance in the directions of higher powers, higher frequencies, and higher 
voltage and current ratings. 


The construction of a power transistor is represented in Fig. 1. 

An important factor in the design of a power transistor is the emitter 
efficiency. In a p-n-p transistor, this is the ratio of the hole current to 
the total current across the emitter p-n junction, when it is positively 
biased in the forward direction. The emitter efficiency should be as 
high as possible. 

The current amplification factor a' normally falls off as the operating 
current is increased, and the resulting non-linearity may be sufficient 
to limit the useful working range of the transistor. This decrease in a' 
depends on the emitter efficiency, which in turn depends on the solid 
solubility in germanium of the additive used for the emitter. A good 
emitter efficiency will be obtained, provided the conductivity of the 
p-type region is much higher than that of the n-type. A high solid 
solubility gives a high conductivity in the p-type region, and therefore 

emitter lead 

metal mounting base 
collector disc 

Fig. 1 — Construction of power transistor 

a high hole current. This in turn leads to a much more constant value 
of a' as the operating current is increased. 

The solid solubilities of p-type additives in germanium increase in 
the order indium, gallium, aluminium and boron. At first, p-n-p tran- 
sistors were made with emitter pellets of pure indium, but with the 

Page 9 


demand for a higher a' at a higher operating current, the emitter 
efficiency was increased by the addition of a small quantity of gallium. 
Even better emitter efficiency (and hence linearity) can be obtained 
by adding a small amount of aluminium to the emitter. 


The base of the OC23 is very narrow and very uniform over a large 
area. The ratio of the diameter of the base to its width is about 80: 1, 
as compared with about 25 : 1 for other germanium alloy-junction 

To obtain this geometry, the penetration of the emitter into the base 
wafer has been kept small, so that the curvature of the edges of the 
emitter junction is much smaller than in transistors with larger emitter 

A special forming process has been developed to give the high- 
quality junctions necessary in this transistor. 

The OC23 is designed and specially tested for driving square-loop 
ferrite computing elements and storage matrices. It provides one-amp 
pulses with a rise time of less than 0-SVsec. 


Silicon transistors at present are made by the p-n-p alloy-junction 
process. Aluminium discs which form the emitter and collector are 
alloyed to opposite faces of a silicon disc (Fig. 2). 

gold ring (for connection 
of base lead) 

Fig. 2 — Construction of a.f. silicon transistor 

Two major advantages result from following what is fundamentally 
the same conception as that used for the OC71. First, basically the 

Page 10 


same considerations apply when designing equipment for silicon as for 
germanium alloy-junction transistors, and equipment designers can 
derive maximum benefit from their experience gained with the ger- 
manium types. Second, the manufacturing experience obtained with 
millions of germanium devices is being used in the large-scale pro- 
duction of the silicon types. 

Silicon transistors combine a number of distinct features : 

(a) low leakage current 

(b) relative freedom from thermal runaway 

(c) higher permissible junction temperature 

(d) higher collector-voltage ratings 

(e) low bottoming voltage 

(/) gain maintained at very low temperatures. 


In alloy-junction transistors, the current is transferred by the diffusion 
of charged particles from the emitter to the collector junction within 
the field-free region of the base. Diffusion is a slow process, which 
severely limits the frequency performance of the transistor. In fact the 

emitter lead 
P-type emitter pellet 
P-type recrystallised layer 
P-type diffused loy< 


s : ^crystallised layer 

-type diffused layer 

collector connection 

P-type germanium crystal 

Fig. 3 — Construction of alloy-diffused transistor 

OC44 and OC42 are made with the narrowest base attainable in the 
alloy- junction technique, and give the best h.f. performance possible 
with this technique. 

In the alloy-diffused construction (Fig. 3), the width of the base is 
reduced to a few thousandths of a millimetre, and an accelerating or 
'drift' field is introduced between the emitter and collector junctions. 

Page 11 


The accelerating field is created by doping the emitter pellet with 
both p- and n-type additives, and heating to a high temperature for a 
carefully controlled time. The n-type additive penetrates the crystal 
more deeply than the p-type, and forms a graded base layer which 
accelerates the holes towards the collector. 

Page 12 



The transistor will now be regarded as a three-electrode device, and 
the relationships between the currents and voltages at the electrodes 
examined. Comparisons will be made with the thermionic valve to 
biing out similarities and dissimilarities. 

The transistor has four main parameters, namely, input voltage and 
current and output voltage and current, and its d.c. performance is 
described completely by a set of four graphs. 

There is more than one way in which the set can be selected, but the 
graphs shown in this chapter are the most convenient. These graphs 
are typical of a small-signal transistor. 

The temperature dependence and the techniques for dealing with it 
will not be considered until later chapters. 


The transistor is usually regarded as a current amplifier, and its 
characteristics are often described in terms of current. 

The base current, which is only a few percent of the emitter current, 
is important because it controls the current in the emitter-collector 
circuit. A similar controlling function is exercised in the valve by the 
control-grid voltage. 

The direct or alternating currents at collector, emitter and base can 
be added up in the usual way. If the direct emitter current is 5mA and 
the collector current 4 -9mA, the base current is lOO^A. 

A definite relationship exists between the currents at the transistor 
electrodes. This relationship is expressed by two ratios, a and a', which 
are characteristic of the transistor. 

a is the general symbol for collector current divided by emitter 
current, and a' for collector current divided by base current. One can 
be obtained from the other, since a' = a/(l— a) and a = a'/(l+a'). If 
the current at one electrode is known, it suffices to know either a or a' 
to find the currents at the other electrodes. 

In practice the values of a and a' to be used may depend on the 
frequency and current. In the present chapter only the simplest case 
will be considered, that of small changes of current at zero frequency 

Page 13 



The input characteristic shows the variation of input current with 
input voltage. When the input is applied to the base, the input charac- 
teristic shows the base current lb plotted against the base voltage Vb 
(Fig. 1). The base voltage is measured relative to the emitter, and 
corresponds to the voltage Vbe between base and emitter in a practical 

When the input electrode is the emitter, the input characteristic 
shows a plot of emitter current I e against emitter voltage V e (Fig. 2). 
It is usually necessary to interpret V e as V e b in practice. 

Input Resistance 

The reciprocal of the gradient of the input characteristic is the input 
resistance. This is typically of the order of 50 to 100Q with the input 

Grounded emitter 
V e = -4-5V 



Fig. 1 — Input characteristic in grounded emitter 

applied to the emitter, and 500Q to lkQ with base input, and is much 
lower than for a valve. 

The input characteristic derives from the characteristic of the forward- 
biased emitter junction. The input resistance is therefore low. The 
characteristic is approximately exponential in shape, as may be pre- 
dicted theoretically for the semiconductor diode. 

The input characteristic is thus quite non-linear. The input resistance 
depends very much on the current at which it is measured. 

The collector voltage is kept constant during the plotting of the 
input characteristic; in a practical circuit, constant collector voltage 
could only arise with the output short-circuited to a.c. (zero load). 

Page 14 


Current Drive 

Since the resistance of the emitter diode changes with the current 
flowing through it, as shown by the non-linearity of the input charac- 
teristic, the transistor is normally current biased and driven from a 

Grounded base 
V c = -4-5V 


Fig. 2 — Input characteristic in grounded base 

current rather than a voltage source. Current drive is achieved by 
using an effective source resistance which is large in comparison with 
the input resistance. 

Input-circuit Distortion 

If the constant impedance of the source is not sufficiently high to 
swamp the varying impedance of the transistor under drive, high 
input-circuit distortion will result. 

When using an oscilloscope to examine the input waveform for 
distortion, the current waveform should be monitored, as often the 
voltage waveform is misleading. 


The transfer characteristic of a valve shows the variation of anode 
current with control-grid voltage, that is, the dependence of the output 
current on the input voltage. The mutual conductance or slope g m of 
the valve corresponds to the gradient at some point of this character- 
istic, and is normally specified in milliamps per volt. A similar quantity 
is sometimes expressed for the transistor; this is the gradient of the 
Ic/Vb characteristic, which is of the order of tens or hundreds of 
milliamps per volt. 

However, if the collector current is plotted against the base current, 
the resulting curve is much more linear than the I c /V b characteristic. 
This characteristic (Fig. 3) is the one normally referred to as the 

Page 15 


transfer characteristic of the transistor in grounded emitter (input 
to base). 

The transfer characteristic in grounded base (input to emitter) shows 
I c plotted against I e (Fig. 4). 

Non-linearity Distortion 

Any non-linearity in the transfer characteristic gives rise to non- 
linearity distortion in the output. This distortion is low for the OC71 
and similar transistors operated at collector currents in the region of 

Grounded emitter 
V c = -4-5V 



Fig. 3 — Transfer characteristic in grounded emitter 

a milliamp or so; however, operation below 0-3mA or above 25mA 
is inadvisable when low distortion is required. 

Current Amplification Factor* with Input to Base (Grounded 

The gradient at any point of the I c /Ib transfer characteristic is a ratio 
and has the dimensions of a pure number. This is the current amplifica- 
tion factor of the transistor with the input applied to the base. It is 
represented by <x' (alpha nought dash). 

The subscript o indicates that <x' applies to zero frequency (d.c). 

From Fig. 3, it can be seen that the base current is 250[xA at a 
collector current of 14mA, the collector voltage being kept constant 
at — 4-5V. The corresponding value of a'<, is therefore 1400/25 = 56, 
the characteristic being linear in the range shown. 

*The term current amplification factor may be applied to transistors, by analogy 
with the voltage amplification factor of a valve. The more usual expression is current 
gain; sometimes this term is qualified as the transistor current gain, to distinguish it 
from the current gain of the stage. 

Page 16 


a' is not greatly affected by collector voltage. However, the I c /Ib 
characteristic is measured at some constant value of collector voltage 
which, for the sake of completeness, is shown on the curve. Constant 

Grounded base 
V c =— 4-5V 


Fig. 4 — Transfer characteristic in grounded base 

collector voltage could only arise in a practical circuit with the output 
short-circuited to a.c. 

The definition of a' Q is given mathematically by 

, _ aicl 

This equation restates in mathematical terms that <x' is a numerical 
quantity, which corresponds to the slope of the I c /Ib curve, and is 
measured at constant collector voltage. 

Current Amplification Factor with Input to Emitter (Grounded 

The gradient of the I c /I e curve (Fig. 4) is the current amplification 
factor in grounded base and is given the symbol <x . The relationship 
is linear in the range shown, and a is 9-8/10 = 0-98. a is defined 
mathematically as 



The output characteristic of a valve consists of a plot of anode current 
against anode voltage for various control-grid voltages. 

The output characteristic of the transistor, which is normally current 
driven, has input current as parameter instead of input voltage. Such a 
graph is shown in Fig. 5 for a transistor in grounded emitter (the 

Page 17 


input current being the base current). The base current is normally 
negative, according to the usual sign convention. 

In a practical circuit, which may contain collector and emitter 
resistors, for example, the collector voltage V c is the collector-emitter 
voltage V ce across the transistor. 

Fig. 6 shows the output characteristic in grounded base. The input 
current is now the emitter current which is plotted as parameter. 

Output Resistance 

At voltages above the knee voltage, which is only about 0-2V in Fig. 5, 
a comparatively large change in collector voltage produces a relatively 

Grounded emitter 




P 1 












Fig. 5 — Output characteristic in grounded emitter 

small change in collector current. The transistor therefore has a high 
output resistance. This can be explained by regarding the collector- 
base junction as a diode biased in the reverse direction. 

The gradient of the output characteristic, because of the way it is 
plotted, has the dimensions of a conductance (current/voltage) which 
is low. The reciprocal of this is the output resistance, which is high. 
The curves apply to constant base current, which would only be 
obtained with a high source resistance in the base circuit. 

Pentode-like Characteristic 

The transistor, which is basically a three-electrode or 'triode' device, 
has an output characteristic like that of a pentode valve in having a 
well-defined knee and a region of high output resistance. The knee 

Page 18 


voltage of the transistor (about 0-2V to 3V) is much lower than that 
of the pentode valve (which may be about 30V). Hence low battery 
voltages can be used for transistors, while still obtaining high efficiency. 

Output Characteristic related to Transfer Characteristic 

If a straight line be drawn across the output characteristic for some 
particular value of collector voltage, and the readings of I c and I b or 
I e replotted, the result is the transfer characteristic for the collector 



ed base 










!". 4- 









-4 -8 


Fig. 6 — Output characteristic in grounded base 

voltage in question. The non-linearity in the transfer characteristic 
therefore appears as an uneven spacing of the curves on the output 
characteristic, for equal changes in input current. 

Collector Leakage Current 


A closer inspection of the output characteristic in grounded emitter 
(e.g. Fig. 5) leads to the conclusion that a finite collector current 
continues to flow, even when the base is open circuit (I b = 0). This 
current is called the collector leakage current in grounded emitter, 
and is commonly given the symbol I' co . Alternatively it may be 
represented by I ceo , where the subscript ce designates a collector- 
emitter current, and the subscript shows that the current in the 
remaining electrode (the base, b) is zero (Fig. 7). 

As is shown by Fig. 5, l' co increases slightly with collector voltage. 

The value of I' co is typically 150[jlA for an OC71 (V c = — 4-5V), 
at an ambient temperature of 25 °C. 

Page 19 



The collector leakage current in grounded base is the base-collector 
current with the emitter open circuit (Fig. 8). The corresponding 
symbol is I co or I C bo. 

Ico is typically 4-5[xA for an OC71 at V c = — 4-5V and at an ambient 
temperature of 25°C. This quantity is too small to be shown on the 
output characteristic in grounded base (Fig. 6). 

Ico is the current flowing in the collector-base diode biased in its 
reverse direction (Fig. 8). The term 'leakage current' arises because 
no current would flow through an ideal diode under similar conditions. 

The full value of leakage current in grounded-emitter circuits 
( = l' c0 ) is only obtained when the base is open circuit. When the 

Fig. 7 — Collector leakage current in grounded emitter 
Fig. 8 — Collector leakage current in grounded base 
Fig. 9 — Emitter leakage current 

base is connected by a resistance to ground (i.e. to emitter), a leakage 
current lying between I'co and a value somewhat higher than I co is 
obtained. The leakage current does not drop to I co when the base is 
short-circuited to ground, because of the small forward bias produced 
across the emitter-base junction by the flow of leakage current through 
the internal base resistance rt,b' . 

Emitter Leakage Current 

If the collector is open circuit as in Fig. 9 (I c = 0), the transistor 
reduces to a base-emitter diode. The current which flows when this 
diode is biased in the inverse direction, with the emitter negative for a 
p-n-p transistor, is the emitter leakage current, I e o or Iebo- 

The emitter leakage current is usually of the same order of magni- 
tude as the collector leakage current in grounded base. Thus for the 
OC71, I e0 is typically 3-5jxA at V e = — 4-5V and at an ambient 
temperature of 25°C. 

Page 20 



The variation of base voltage (base-emitter voltage) with collector 
voltage may be shown by means of a feedback characteristic. This curve 





Itv« l20yA 




Fig. 10 — Feedback characteristic in grounded emitter 

(Fig. 10) is not of any great practical value, and is no longer given in the 
published data. 

The effect of feedback within the transistor is extremely important, 
and not directly comparable with anything that occurs in a valve. 
This point will be taken up again when discussing small-signal charac- 
teristics and equivalent circuits. 


The static characteristic curves are normally given for 

(a) emitter grounded (input to base, output from collector) 

(b) base grounded (input to emitter, output from collector). 

The curves are not quoted with reference to grounded collector (input 
to base, output from emitter). 

The information contained in the curves applies whichever electrodes 
are used for the input and output. 

For example, the output characteristic in grounded emitter expresses 
a relationship between I c , V e and lb , which is still valid in the other 
configurations, provided V c is interpreted as V ee . Similarly, the I c /Ib 
curve is true for all configurations, although the term transfer charac- 
teristic is appropriate to this curve only when the transistor is in 
grounded emitter. 

Page 21 



(a) The thermionic valve has a comparatively high input impedance 
and is voltage driven. The transistor, on the other hand, has a 
comparatively low input resistance (impedance), and normally 
must be current driven from a high source impedance. 

(b) The characteristics of the transistor are often expressed in terms of 
input current rather than input voltage. 

(c) The output characteristic of the 'triode' transistor resembles that 
of a pentode valve in having a definite knee and a region of high 
output resistance. The knee voltage of the transistor is much lower 
than that of the pentode valve, and low supply voltages can be 
used while retaining high efficiency. 

(d) When one of the electrodes is open circuit, the transistor does not 
reduce to an ideal diode, because leakage current flows through 
the reverse-biased diode. 

(<?) Transistor characteristics are normally given for grounded emitter 
and grounded base. This method of presentation is more extensive 
than that adopted in valve data. 

(/) One of the peculiarities of the transistor is its internal feedback. 

Page 22 



The static characteristic curves described in the previous chapter are 
typical of those normally given. Such curves are useful primarily for 
choosing a working point, and in particular for the design of a.f. 
output stages. The curves are less useful for the design of small-signal 
a.f. stages. 


The most profitable way of approaching the a.c. performance of the 
transistor is to treat the device as a four-terminal (or four-pole) 'black 
box' (Fig. 1). The box has a pair of input and a pair of output terminals. 


Fig. 1 — A.C. quantities of 'black box* in grounded base 

Relationships between the signal voltages and currents measured at 
the input or output of the black box are called four-pole character- 
istics or parameters. 

The 'black box' treatment is not peculiar to transistors. It is a 
general method which can be applied to any electrical network. 

Small-signal characteristics are published for a.f. transistors in the 
form of 'h' and 'modified z' systems. These parameters are suitable 
for calculating performance at audio frequencies or, more precisely, 
at frequencies low in comparison with the 'cut-off frequency'. 

For some h.f. transistors the four-pole characteristics are quoted 
as a set of four admittances (y parameters). 


One set of small-signal a.f. characteristics is formed by the slopes of 
the static characteristic curves at the working point. The slopes cannot 

Page 23 


be found with sufficient accuracy from the graphs, and so they are 
given for one or two nominal working points. 

The symbol for these characteristics is an h which is modified by 
numbers i and 2 in subscript. A subscript 1 refers to a voltage or 


Slope =h2i 


Slope =h22 

V 2 


Fig. 2 — Notation for h-parameter subscripts 

current measured at the input terminals, while a subscript 2 denotes 
a voltage or current at the output terminals (Fig. 2). 

Capital letters represent direct voltage and current at the input 
electrode (Vi , Ii) and at the output electrode (V 2 , I 2 ). Small letters 
will represent a.c. quantities at the input electrode (vi , ii) and output 
electrode (V2 , i2>. In practice vi , ii and V2 , i2 represent small 

Characteristics in the h system are defined as follows : 
hn — vi/ii = slope of input characteristic 

= input impedance for constant output voltage 

1121 == i2/ii = slope of transfer characteristic 

= current amplification factor for constant output 

1122 = i2/v2 = slope of output characteristic 

= output admittance for constant input current 
( = reciprocal of output impedance) 
hi2 = V1/V2 = slope of feedback characteristic 

= voltage-feedback ratio for constant input current. 
Constant output voltage means that V2 = and that the output (load) 
is short-circuited to a.c. (zero load); for example, a large capacitance 
may be connected across the output terminals. Constant input current 
means that ii = and that the input is open-circuited to a.c, as for 
example by including a large series resistance or inductance. Thus it 
is usual to see the characteristics expressed as: 

hn = input impedance with output short-circuited to a.c. 

1121 = forward current transfer with output short-circuited to a.c. 

1122 = output admittance with input open-circuited to a.c. 

hi2 = reverse voltage transfer with input open-circuited to a.c. 

Page 24 


hn is measured in ohms or kilohms, I121 is a ratio (that is, a pure 
number), J122 is measured in reciprocal ohms (mhos or micro-mhos) 
and hi2 again is a ratio. Because of their different dimensions, these 
parameters are referred to as hybrid, whence the symbol h. 

Similar quantities can be defined for grounded emitter from the 
appropriate set of characteristic curves, the symbols then being primed, 
thus: h'n, h'21, h'22 and h'12. The values of the primed and unprimed 
quantities are normally different, corresponding to the different values 
measured in grounded emitter and grounded base. 

h'| 2 


a 3m 



h 22 


h 2l 

h', ? 





h'| 2 

Vc = 


, h' 22 

h' 2 i 


h' 22 

-I -2 -5 HO 


12 5 10 


Fig. 3 — Variation of h-parameters with working point, for OC71 in grounded emitter 

The values of the h parameters as given by the slopes of the static 
characteristic curves would only apply to zero frequency. In practice, 
the parameters are measured at a representative audio frequency 
(lOOOc/s), using special equipment. 

The h parameters can also be defined from the equations: 

vi = h 1 iii+h 12 v 2 ...(1) 


i 2 = h 2 iii+h 2 2V2. ...(2) 

The values of the characteristics also depend on the working point. 

Fig. 3 for the OC71 in grounded emitter enables the characteristics 

for other working points to be derived from the values at — 2 V, 3mA. 

Performance equations based on the h-parameters are given in 
Table 1, at the end of the chapter (p. 31). 

This system of characteristics shows to particular advantage when 
it is required to calculate the performance of circuits containing 
several stages, feedback networks, etc. The characteristics of the transis- 
tor can be expressed in the form of an h matrix, a matrix merely being 
a special notation for writing down coefficients. Thus Eqs. 1 and 2 
can be re-written as 

("vil = Thii hi 2 "| I" ill 
L »2j ~~ U21 I122J |v2j 

Page 25 


The characteristics of any network, such as a feedback path for instance, 
can also be expressed in matrix form. The performance of a circuit 
containing transistors and other networks can be calculated by mani- 
pulating the matrices according to certain rules (matrix algebra)*. 


The subscripts 1 and 2 constitute a general notation for distinguishing 
quantities measured at the input and output terminals. The voltages 
and currents can be combined in the appropriate ways to give resist- 
ance and impedance (r and z), admittance (y) and conductance (g). 


A set of four parameters which all have the dimensions of impedance 
can be written out using the general notation, thus: 

zn = vi/ii = input impedance with output open-circuited to a.c. 
Z21 = V2/ii = forward transfer impedance with output open-circuited 

to a.c. 
Z22 = V2/i2 = output impedance with input open-circuited to a.c. 
Z12 = Vi/i2 = reverse transfer impedance with input open-circuited 
to a.c. 


Equations based on the h system (Table 1) do not indicate clearly the 
characteristic values and the trends in performance as the quantities 
are varied. A further set of a.f. characteristics has therefore been 
produced. These are more suitable for elementary circuits which do 
not include a.c. feedback and other networks. 

The system consists of five characteristics which are again defined 
from the 'black box'. They are (for grounded base): 

forward current transfer a at constant collector voltage (output 
short-circuited to a.c.) 

input impedance zi n at constant collector voltage (output short- 
circuited to a.c.) 

input impedance zn at constant collector current (output open- 
circuited to a.c.) 

output impedance z ou t for constant input voltage (input short- 
circuited to a.c.) 

output impedance Z22 for constant input current (input open- 
circuited to a.c.) 

* Matrix methods for circuit calculations are described in Principles of Transistor 
Circuits, edited by R. F. Shea. Wiley (New York). Chapman and Hall (London). 

Page 26 


From these definitions, three of the five characteristics will be 

seen to 

be related directly to the h characteristics, thus : 

a = 1 1121 1 

Zin = hii 



Z22 =7— • 


The other two, 

zii and z ut , are related to the h system as 

follows : 

zn = hii— h2i.— — 





" A ' 

hnh22— h2ihi2 


A = hnh22— h2ihi2 . 

A set of five characteristics is defined in a similar way for grounded 
emitter, the symbols being primed, thus: a', z'm, z'n, z'out and z'22. 

The equations using this system (Table 2) consist of a simple fixed 
term, dependent on the transistor only, multiplied by a term which 
varies in a fairly simple manner with the circuit values. 

One of the merits of this system is that it shows the interdependence 
of the input and output impedances. In an actual circuit, the input 
impedance lies between the values with open-circuit and short-circuit 
output. The output impedance lies between the two extreme values 
which correspond to open-circuit and short-circuit input. The inter- 
dependence of the input and output impedances is the effect of internal 
feedback, which is allowed for in the h system by the voltage-feedback 
factor hi2. 

As can be seen from the definitions, 

z'n = zn and z'out = z OH t. 

Other useful relationships are: 

z'in = Zin(l+a') ~ a'zinj 
Z22 _. Z22 

z 22 = 

1+a' a' 

Zin Zout . j z In z out 

Zn Z22 ' z'n z'22 

Page 27 


Dynamic Performance 

The modified z characteristics are particularly useful for calculating 
the dynamic performance of individual transistor stages. 

The equations to be applied depend on the relative magnitude of 
the load resistance Rl . Certain basic equations can be derived for an 
Rl which is of the order of the output impedance (say, 30kQ). When Rl 
is low, only lkO or so, the equations can be simplified. 

In fact the choice of equations is decided by the coupling. The 
value of Rl which will give maximum gain is high, and can normally 
only be provided by transformer coupling. 

When RC coupling is favoured, the load is virtually formed by the 
input impedance of the following stage, which is low and effectively 
shorts the collector resistor to a.c. 

Hence the calculations are simpler to perform for RC coupling. 


Power gain is a maximum when the load resistance is Rl = VXzWout) 
and the source resistance is R s = V(z'iiz'in). 

Maximum gain occurs with these optimum values and is given by 
Matched Power Gain = 

Z 22- 


(Similar equations apply in grounded base, using unprimed quantities.) 

As an example, consider a typical OC75 in grounded-emitter con- 
nection. The following values of the parameters are for a working 
point of — 2V, 3mA. With z' 22 = 7-8kQ and z' ou t = 14kD, the 
optimum load resistance is V(7-8xl4) = 10-5kQ. With z'n = 720Q 
and z'in = 1 -3kO, the optimum source resistance is 

V(720x 1300) = 9700. 
The matched power gain is 

| v720+Vl34 2x7800= ' 5 ' 90 ° = 42dB - 

where the power gain in dB is ten times the logarithm to the base ten 
of the numerical power gain; thus 101ogi 15,900 = 10 X4-2014 = 42dB. 
Similar calculations for grounded base, using the unprimed quantities 
a, zn , zm and z 2 2 , show that the maximum gain available is about 
29dB in this configuration. 

For multistage amplifiers, matching can easily be achieved with 
interstage coupling transformers. A stepdown ratio is required. 

Page 28 



With RC coupling, the collector resistor is normally several kQ's, and 
the input resistance of the following stage only about 700Q. Thus the 
output is more or less shorted to a.c. The current gain Ai and the input 
impedance of the grounded-emitter stage are approximately equal to 
the small-signal parameters, a' and z'm . 
The voltage gain A v of the stage is 

Z in 

where R L , the load resistance, is substantially equal to the input 
impedance of the following stage. 

The power gain A w is 

A w = AixAv 


For identical stages in cascade, R L = z' ln and A w = (a') 2 , so that 
for a typical a' of 41 (for the OC71), 

A w = (41)2 = 1680 = 32dB. 

In a practical RC-coupled stage, the power gain will always be less 
than (a') 2 because: 

(a) there is considerable loss in the coupling network (collector 
resistor, coupling capacitor, and biasing components); 

(b) the factor Rl/z'id is less than one, because succeeding stages are 
operated at higher currents to accommodate the increased signal 
swing, and the input impedances of the stages become progres- 
sively lower. 

The power gain of an OC71 stage with RC coupling is typically about 
26 to 30dB. 


The low-frequency T network is described at greater length in a later 
chapter. Performance equations based on the T-network parameters 
are given in Table 3 for grounded-emitter operation. For the present 
purpose, the notation of these parameters is sufficiently explained by 
Fig. 4, where one version of the equivalent circuit includes a voltage 
generator i e r m , and the other a current generator ai e . The equations 
can be considerably simplified by neglecting r m in comparison with 
r e and r c . 

The chief disadvantage of the T-network parameters, as compared 
with those defined in the h and modified z systems, is that they lead to 

Page 29 


cumbersome expressions. Furthermore, a different equation is required 
for each of the three circuit configurations. The parameters them- 
selves, however, have the same value for each configuration. 

The parameters, which depend to some extent on the operating 

o — A'W 1 — ( <\( 1 (/V^A— • 

Fig. A — Low-frequency T network 

conditions, are useful up to a point for understanding large-signal 
performance, if average values are employed. 

Tables 4 and 5 show how to evaluate the h and modified z para- 
meters from the T-network parameters. 


The y parameters form another set of four-pole characteristics. The 
subscripts are used as for the h parameters to denote quantities 
measured at the input and output terminals. Similarly, a dash or prime 


y 11 = 
(g'11 + jwc'ii) 

V 2V12- 
(g' t 2 + jwc'i2)v2 

y'22 = 

Fig. 5 — y-parameters and their components 

indicates values for grounded-emitter operation. Unlike the h para- 
meters, however, the y parameters all have the same dimensions, 
namely, admittance. Thus in grounded emitter: 

y'n = ii/vi = input admittance with output short-circuited to a.c. 
y' 21 = i 2 /vi = forward transfer admittance with output short- 
circuited to a.c. 
y'22 = i2/v2 = output admittance with input short-circuited to a.c. 
y'12 = 11/V2 = reverse transfer admittance with input short-circuited 
to a.c. 



Values of the y parameters are given in the published data for r.f. 
transistors such as the OC170. The input, output and reverse transfer 
admittances are resolved into a conductance and a capacitance (Fig. 5). 
The three conductances increase with frequency and the capacitances 

The phase shift within the transistor is given as $'21 , the phase angle 
of the forward transfer admittance. 


Grounded Base* 

G L = 1/R L 
Input Resistance 

Output Resistance 

Current Gain 
Voltage Gain 
Power Gain 

A = hnh22— hi2h2i 


1122 + Gl 




h22 + G L 

A+hnG L 
(h 2 i) 2 G L 

(h 22 +G L )(A+hiiG L ) 
* For grounded emitter use the same equations with primed quantities. 



Grounded Base* 

Input Resistance 
Output Resistance 
Current Gain 
Voltage Gain 
Power Gain 





Rs + Zln 


R L +Z22 

0CZ22 R L 

Zll (R L + Zout) 
(«Z22) 2 R L 

Zll (R L +Z22)(R L +Zout) 

"For grounded emitter use the same equations with primed quantities. 

Page 31 




Grounded Emitter 

Input Resistance 
Output Resistance 
Voltage Gain 
Power Gainf 

r b +r e + 

r e (r m — r e ) 

r e +r c — r m + 

Rt+re+re— r m 
r e (r m — r e ) 

Rs+rb+r e 
(r e — r m )RL 

(Rs+rb+r e )(RL+re+r c — r m )+r e (rm— r e ) 
4RLR s (r m -re) 2 

t(Rs+r b +re)(RL+re+ro-rm)+r e (rm-re)] 2 

*These equations can be simplified by neglecting r m in comparison with r c and r P 
tPower output divided by max. power obtainable from generator 








hn = r e +(l — a)n> 

h'ii = (l + « / )hn 

h*n = li'n 

h2i = a 

h'2i = a' 

— h'21 = l+h'21 

b.22 = l/r 

h'22 = (l + aOh.22 

h"B2 = h'22 

hi2 = rb/r c 

h'i2 = h'22r e 

h"is = l/a+h'12) 




Grounded Base 

a = r m /rc 
zn = r e +rb 

zm = r e +(l — a)r b 

Z22 = r c 


\ T e +Ib) 

Grounded Emitter 

V-' = Tml(To—Tm) 
Z'll = Te+Ih 

z'm = rt,+r e (l + «') 
z'22 = r c /(l + «0 

Page 32 



The three basic circuit arrangements of the transistor will be con- 
sidered in this chapter from the a.c. point of view. The modified z 
parameters, described in the previous chapter, will be used to compare 
the performance, as the equations based on these parameters show 
most clearly the effects of the load and source resistances. 

For the sake of comparison, the basic circuits of the triode valve are 
given in Figs. 1, 2 and 3. The first is the grounded-grid circuit, in whioh 

Fig. 1 — Grounded grid 
Fig. 2 — Grounded cathode 
Fig. 3 — Cathode follower 

the input is applied to the cathode. In the second, more common 
arrangement, the input is applied to the grid and the cathode usually 
grounded to a.c. Finally there is the cathode follower. 

There are similarly three a.c. circuit configurations for the transistor 
(Figs. 4, 5 and 6). They are usually known as grounded base, grounded 
emitter and grounded collector, according to which electrode is com- 
mon to the input and output circuits. 

Table 1, at the end of the chapter (p. 43), summarises the character- 
istics of the three transistor configurations, and is useful when choosing 
the one required for any particular purpose. 

Page 33 



The grounded-base or common-base circuit (Fig. 4) gives the best 
illustration of how a transistor works, though it is not the most fre- 
quently used. The input is applied to the emitter and the output taken 






39< * 

2V-i- + 

Fig. A — Grounded base 

from the collector. The base is common to the input and output cir- 
cuits, and is normally grounded to a.c. 

Input Impedance 

The input impedance of the transistor in grounded base is 

Z in = Zii. RL + Zout . 

From the definitions of the modified z parameters, the input impedance 
lies between the extreme values of zn (output open circuit) and zm 
(output shorted). Thus if in the above expression Rl-> oo, then 

Zin-> zn 
and if R L = 0, 

rj _ ZllZ0Ut _ 

Zjin — ' — Zin 


since z ou t/z22 = zm/zn . 

For a typical OC71 at a working point of — 2 V, 1mA, 
zm = 350 and zu = 720O. 
The input impedance therefore varies widely with the load resistance, 
and in this particular example, by as much as 20:1. 

Output Impedance 

The output impedance in grounded base is 

7 _ rr Rs+Zin 

Amt = Z22 • - — ■ • 


Page 34 


The extreme value of output impedance is z 2 2 for R s -> oo (input open 
circuit). For R s = (input shorted), the extreme value is 

Z22Zin _ ~ , 

The variation of output impedance with source resistance is the 
same as the variation of input impedance with load resistance, since 

Zin Zout 

Zll Z 2 2 

For the typical OC71 and a working point of — 2V, 1mA, 
Z22 = 1 -0MO and z ou t = 50kO. 
The variation is therefore 20:1, as expected. 

The ratio of output to input impedance is high, thus 

^ = ^ ~ 1400. 

Zin Zn 

Current Gain 

The current gain in grounded base, in terms of the modified z para- 
meters, is: 


Ai = a. 

RL + Z22 

If the load resistance Rl is made equal to zero (short-circuited output), 
the current gain becomes equal to a. 

As Rl is increased, the current gain becomes progressively smaller. 
When Rl is infinite (open-circuited output), Ai is zero. 

For an alloy-junction transistor, a is slightly less than one. The 
current gain in grounded base is therefore always less than one. 

Current Amplification Factor 

a is the current amplification factor of the transistor in grounded base. 
This is the maximum theoretical current gain. 

The condition Ai = a when Rl = (output shorted to a.c, V2 = 0) 
agrees with the definition of a given in the previous chapter. 

Voltage Gain 

The voltage gain in grounded base is 

az22 Rl 

A v = 

Zll Rli+Zout 

The factor Rl/(Rl+z uO approaches one as Rl is increased; the 
voltage gain increases as Rl increases, and reaches a theoretical 
maximum when Rl is infinite. A v is zero when Rl is zero. 

Page 35 


Although a is typically about 0-98, the ratio of the output impedance 
to input impedance (z 2 2/zn) is high, and so is the possible voltage gain. 

Voltage Amplification Factor 

The maximum voltage gain, with infinite load (R L -> oo) is 

/V2\ 0CZ22 

^21 = I — Ii2 = ■= — — • 
\Vl/ Z11 

[X2i is the voltage amplification factor, and is a function of the tran- 
sistor only. 

The definition of jjl 2 i is similar to that of the voltage amplifica- 
tion factor [x of the triode valve. However, for transistors, a \i without 
subscript is usually reserved for fjn 2 , the voltage feedback factor 
( ^ h i2 ). 

For a typical OC71 operating at — 2V, 1mA: 

a = 0-976, Z22 = 1-0MQ, and zn = 720a 
Hence the voltage amplification factor (or maximum voltage gain) is 

az 22 0-976x106 

^21 = — = ^^ — ^ 1400. 

zn 720 

Power Gain 

The power gain A w is the product of the current gain and the voltage 
gain, so that in grounded base 
A w = AiXAv 

= (az 22 ) 2 Rl 

Zll (RL+Z22)(RL+Z ut) 

The grounded-base circuit derives its power gain from its high 
voltage gain. 

The power gain is a maximum when 

Rl = optimum Z ou t = Vfezout), 
as can be shown by differentiating the expression for A w with respect 
toR L . 

The source resistance which gives the optimum output impedance 
is found from 

R s = optimum Zm = VtenZin). 

With the above values of Rl and R s , the matched power gain is 

A w max 

= Z22J- 

For a typical OC71 at a working point of — 2V, 1mA, the modified 
z parameters in grounded base are: 

Page 36 


a =» 0-976; z ln = 35ft; zn = 720Q; z ou t = 50kQ; andz 22 = 1-0MQ. 
Consequently the matched power gain at this working point is 
/ 0-976 \ 2 
106X ( V720 + V35 J "WO "2MB. 

This power gain is obtained with 

Rl = V(z22Z ou t) = VC5 X lOio) = 224kQ 
and R 8 = Vfcnzin) = V(720x35) = 160Q, 

The maximum power gain is only obtainable with transformer 
coupling, a step-down ratio being required of approximately 

/Rl /224 

Vr7 = 7^16=^1400= 37, 

when feeding into an identically similar stage. 


Grounded-emitter connection (Fig. 5) is the most commonly used of 
the three configurations. The input is applied to the base, the output 
is taken from the collector, and the emitter is common to the input and 

Fig. 5 — Grounded emitter 

output. It is sometimes called common-emitter connection, because 
the emitter is not necessarily grounded to a.c. 

Using the modified z parameters, the performance equations are 
exactly the same as for grounded base, except that the quantities are 
dashed or primed to show that they are measured in grounded emitter. 
The equations will be repeated in their primed form, so that they can 
be associated with the values of the parameters in grounded emitter. 

Input Impedance 

The input impedance in grounded emitter is given by : 

Page 37 

Putting Rlh> oo, 
and putting Rl = 0, 


7/ 1 n = * , ll- RL+Z/ ° ut 

Z'ln -> z'n 

■^ in — ; — Z in , 

Z 22 

as required by the definitions. 
For a typical OC71 at — 2V, 1mA: 

a' = 41, z'n = 720Q, and z' ln = l-45kQ. 

There is therefore comparatively little variation in input impedance 
with load resistance in grounded emitter (z'm/z'ii = 2). 

From the above values, it can be confirmed that z'n = zn and, 
within the errors of measurement, z'i n = zm(l +a'). 

Output Impedance 

The output impedance in grounded emitter is 

Z' ou t = z > 22 . Rs + z ' ia > 

which lies between the extreme values z'22 (Rs -» °o) and 

z'22Z'in _ , 

; Z out 

Z 11 

(R s = 0), as required by the definitions. 

For the typical OC71 as before, z'22 is 25kQ and z' ou t is 50k£2. 
Again the variation is 2:1. 

From the above values, it can be confirmed that z'out = z ou t and 
z'22 = z 2 2/(l + a'). 

The ratio of output to input impedance is medium, thus 

Z out Z 22 ^ t^. 

Z'in Z'n 

Current Gain and Current Amplification Factor a' 

The current gain in grounded emitter is 

Ai= a'. : 

z 22 


As before, the current gain is a maximum when Rl is zero, and 
decreases to zero as Rl is increased to infinity. 

The maximum theoretical current gain a', which is typically 40 to 
50 for the OC71, is much higher than in grounded base, a' is the 

Page 38 


current amplification factor of the transistor in grounded emitter. 

The current-gain equation agrees with the earlier definitions of a', 
in that Ai = a' only when Rl = 0. 

Voltage Gain and Voltage Amplification Factor 

The voltage gain in grounded emitter is 

A _ a'z'22 r l 

Z'n R-L+Z'out 

The maximum voltage gain, when Rl approaches infinity, is 



which is the voltage amplification factor in grounded emitter. This 
value is exactly the same as in grounded base, since 
z'n = zn and a'z'22 = az22. 

Thus the maximum voltage gain ( = (Z21) in grounded emitter, as 
in grounded base, is approximately equal to 1400 for the example 

Power Gain 

The power gain in grounded emitter is much higher than in grounded 
base because, although the voltage gain is the same, the current gain 
is increased. The power gain is calculated from 

a - ( a ' z/ 22> 2 Rl 

Z 11 (Rl+Z 22)(Rl+Z out) 

For an OC71 at a working point of — 2V, 1mA as before, the para- 
meters are: 

a' = 41; z' ln = l-45kQ; z'n = 720O; z' ou t = 50kO; 
and z'22 = 25W2. 
The maximum power gain is 


''■ U-u+Vz'J ' = ^H 

= 40dB. 
This gain is obtained with: 

Rl = V(z'2 2 z'out) = V(25 X50) = 35kQ 

R s = V(z'uz'in) = v"(720xl450) = l-02kO 

The transformer turns ratio is approximately 

VsrVi^ 34 - 3 - 5 - 9 ' 

when feeding into an identically similar stage. 

Page 39 



Grounded-collector connection (Fig. 6) corresponds to the 'cathode 
follower' valve circuit. It is sometimes for this reason referred to as 
the 'emitter follower'. It is also referred to as common collector. 

The input is applied to the base, the output is taken from across the 
load in the emitter, and the collector is common to the input and output. 

The current amplification factor for small signals, which is given the 
symbol a", is approximately equal to a'. The exact relationship is 

<x = 1+a . 

In this configuration, the voltage gain cannot exceed one and the 
power gain is lowest. 

There is high input and low output impedance, and these impedances 
are more dependent on the load and source resistances than in the 

220 S 



— Il^> 


4-7 < 



V ce *»2V I c ^0-5mA 

Fig. 6 — Grounded collector 

other configurations. Over a wide range of values the grounded- 
collector circuit can be considered as an impedance changer, the input 
impedance being approximately ok'Rl and the output impedance R 8 /a'. 

The performance will not be calculated for grounded collector, as 
this circuit is little used for small-signal stages. 


The performance of the various configurations can be compared by 
taking the OC71 as an example. Table 2, at the end of the chapter 
(p. 43), summarises the typical performance of the OC71 for a working 
point of — 2V, 1mA. It is another version of Table 1 into which values 
have been inserted. 

Maximum Current Gain 

The current amplification factor in grounded emitter, a', is 41 and the 
value in grounded base, a, is calculated as 0-976 from 


Page 40 

or explicitly 



1 + a' 

For grounded collector a" = 1+a' — 42. 

The typical value of a' becomes 47 when the collector current is 
increased to 3mA. 

Matched Input and Output Impedances 

The input and output impedances are reduced appreciably when the 
collector current is increased from 1mA to, say, 3mA. Thus at 3mA, 
z'in = 800a, z'n = 500Q, z' ou t = 21kQ and z' 22 = 12-5kQ. The 
matched input impedance is then 630Q and the matched output 
impedance 16-2kQ. 

Matched Power Gain 

The power gain is highest in grounded emitter; useful power gain is 
also available in grounded base, but in grounded collector the power 
gain is rather low. 

For most purposes the power gain is of prime importance when 
choosing the circuit configuration, and the grounded-emitter configura- 
tion is therefore most frequently used, and grounded collector least. 

In none of the configurations is the power gain critically dependent 
on load resistance. In grounded emitter the reduction from 40dB to 
37dB takes place at about 8kO at one extreme and 150kQ at the other. 
The gain in grounded base is reduced from 29dB to 26dB at about 
50kQ and 1MQ. In grounded collector the 16dB gain becomes 13dB 
at about 30O and 20kQ. 

The matched power gain at 3mA is still 40dB, as the increase in a' 
and the fall in input impedance compensate for the fall in output 

Cut-off Frequency 

The current amplification factor of the transistor falls as the frequency 
is increased. The point at which the fall-off becomes pronounced 
depends mainly on the type of transistor and also on the circuit 

The cut-off frequency is the point at which the current amplification 
factor falls to 3dB below the low-frequency or 'zero frequency' value; 
3dB corresponds to 1/V2, that is, 0-707. The symbol for the cut-off 
frequency is f a for grounded base and f ' a for grounded emitter. 

Grounded-base connection gives the best h.f. performance, having 
a higher cut-off frequency (fj than in grounded emitter (f' a ) and 

Page 41 


grounded collector ( ~ f' a ). The OC71 illustrated in Table 2 is only 
an a.f. transistor. For the OC45, f a is typically 6Mc/s, and for the 
OC44, 15Mc/s. 

The maximum frequency at which the transistor will continue to 
give satisfactory performance in any circuit is not necessarily limited 
to the cut-off frequency. Apart from spread in transistor characteristics, 
which means that any given transistor may have a better or worse 
performance than the typical, the choice of a reduction of 3dB to 
define the cut-off frequency is arbitrary. Nevertheless, the cut-off fre- 
quency usually gives a good indication of the suitability or otherwise 
of a transistor for h.f. performance, and is Useful for the purpose of 


A transistor in grounded base will not give useful voltage gain, if it is 
RC coupled into another grounded-base stage. Since the current gain 
is less than one, there is no useful power gain. 

The low input impedance of the following stage virtually shorts the 
collector coupling resistor to a.c. Hence in the expression for voltage 
gain we can write Rl = zi n . Then since a ~ 1, 

Z22 Zm 

A v 

Zil Zin+Zo U t 
z out 

< 1. 

For multistage amplification using RC coupling, the grounded-base 
circuit must be combined with grounded-emitter or grounded-collector 

The above objection does not apply to grounded emitter, because 
with an a' of 40 or more, both the current and voltage gain are greater 
than one. 

With transformer coupling there is no restriction on the cascading 
of grounded-base stages. The impedance ratio in grounded emitter, 
however, leads to more convenient turns ratios, apart from the higher 
matched power gain in this configuration. 


The grounded-emitter configuration gives the highest power gain and 
is normally used in a.f. amplifiers. 

Grounded-base stages are not often used at a.f., but sometimes the 
low input and high output impedances are of value. The low input 
impedance is useful for pre-amplifiers for use with moving-coil micro- 
Page 42 


phones, and the high output impedance for feeding into valve ampli- 

The grounded-collector configuration is useful where high input 
and/or low output impedances are important. It is used in buffer stages 
and can sometimes replace a transformer. 

Despite the higher cut-off frequency in grounded base, this mode of 
connection is not necessarily preferred at high frequencies. Practical 
i.f. amplifiers, for example, can be designed for medium and long 
waves using the OC45 in grounded emitter. 

Grounded-base connection is nevertheless frequently exploited in 
oscillator and switching circuits. 





Current gain 

~ 1 



Voltage gain 



~ 1 

Input impedance 




Output impedance 




Power gain 




Cut-off frequency 



on R L 

Voltage phase shift at low 


~ Zero 

~ 180° 

~ Zero 

Collector Voltage -2V, Collector Current 1mA 

Max. current gain (Rl =0) 
Max. voltage gain (Rl -> oo) 
Matched input impedance (kO) 
Matched output impedance (kft) 
Matched power gain (dB) 
Cut-off frequency (kc/s) 






















Page 43 



V c = — 2V, l c = 3mA; zu = z'u = 590fi, 

Zout = z'out = 16kQ, a = 0-982, a' = 56; 

zu, -= 16Q, Z22 = 590kQ; z'm — 910fl, z' 2 2 = 10-4kQ 

g 300 









Pow»r gain A w (dB), voltage gain A v and current gain Ai 
Page 44 



In Chapters 4 and 5 the transistor has been treated on a four-pole basis. 
Even the static characteristic curves described in Chapter 3 can be 
regarded as being derived from a black box. 

This treatment has the advantage of allowing the performance of the 
transistor in circuit to be calculated, without making any assumptions 
about the interior mechanism of the device. 

For a more complete grasp of transistor operation, and in particular 
in order to understand the biasing arrangements to be described in 
Chapter 7, it is necessary to regard the transistor as something more 
than a set of input and output terminals. Fortunately all the extra 
information required can be derived by treating the transistor as a 
simple current network. 


Use has already been made of the example of a transistor operating at 
an emitter current of 1mA; if the collector current is 0-98mA, then the 
base current must be equal to the difference, 02mA. Similar calcula- 
tions are made on thermionic valves, the screen-grid current of a 
pentode, for example, normally being calculated from the known 
currents at the other electrodes. 

Several important relationships can be derived from this principle, 
provided it is expressed in a form which is easier to manipulate. 

The direct current flowing into the transistor is equal to that flowing 
out, or in symbols 

le-flb+lc = 0. ...(1) 

This equation is a restatement of Kirchhoff's law, which gives, for a 
three-terminal network: 

I1+I2+I3 = 0. 
A similar equation holds for alternating currents, namely: 

ie+ib+ic = 0. ...(la) 

The usual sign convention is followed, according to which current 
flowing from + to — is considered positive. In practice it is usually 
only necessary to consider the sign (direction) of the base current. 

Page 45 



The relationship between the various currents remains the same 
irrespective of the transistor configuration. The difference between a 
and a' arises solely from the choice of input electrode, and a definite 
numerical relationship exists between them. 

From the definitions of a and a', where no regard is paid to the 
direction of current flow, 

a _ Sib 

a' ~~ Si e 

Also, Eq. la can be written numerically as 

ib = i« — ic , 
so that 

Sit, — Si e — Si c . 

« Sj b _ Si c _ . 
a' "~~ Si e 8i e 

Rearranging this equation in its more usual form gives 


1 — a 

Note that 

1+a' = and a = 

1— a 1+a' 


Consider the example given earlier, where the emitter current is 1mA. 
The collector current is about 0-98mA, and the base current is equal 
numeiically to the difference between these, namely, 02mA. 

The signs of these currents will be as follows, for a p-n-p transistor : 
Ie will be positive because it flows from + to — into the transistor: 
Ic will be negative because it flows from + to — out of the transistor ; 
and lb will therefore be negative because, from Eq. 1 

1.0-0-98-002 = 0. 

Thus at normal operating currents and temperatures lb is negative, 
and flows from + to — , out of the transistor. 

To understand the operation of the transistor, it is necessary to 
consider what happens at very low currents. In fact, to take the extreme 
case, consider the emitter open-circuited (I e = 0). A leakage current 
Ico then flows from base to collector. The base current in this condition 

Page 46 


flows from + to — into the transistor, and is positive. 

As the emitter current is increased from zero, the base current 
therefore reverses at some point from positive to negative. 

This effect can be explained by means of Fig. 1, which shows a 
plot of collector current against emitter current. The full line shows 
the behaviour of a normal transistor. When I e = 0, I c = I C o , as 
required. As the emitter current is increased by an amount AT e , the 

B2 B| 


Fig. 1 — Relation of transfer characteristic in grounded base to transistor 


collector current increases by the fraction AI C = a x AI e . The full line 
has a slope of a < 1. (a is the value of a for large current changes.) 

The broken line is drawn at an angle of 45° to the two axes and 
represents the condition in which the collector and emitter currents 
are equal. This line would correspond to a transistor in which a = 1 
and Ico = 0. 

At the point where the two lines intersect, I c = I e and Iu = 0. 
When lb = 0, the base is open circuit, and the collector and emitter 
currents are both equal to r co , the collector leakage current in 
grounded emitter. 

Now examine the point Ai on the full line (a < 1) at a current 
Ic > I' co- The collector current at this point is AiBi and the emitter 
current is CiBi. The base current is equal to the difference AiCi. 
In this region the emitter current is numerically greater than the 
collector current, and as I e is positive, the base current is of the same 
sign as I c , that is, negative (Eq. 1). 

Page 47 


The point A2 lies in the region where I c < l'co- The collector 
current is A 2 B 2 , the emitter current C2B2 , and the base current 
equal to the difference A2C2 . Here the collector current is numerically 
greater than the emitter current, and the base current is positive in 
sign, like I e . 

The transistor can only be operated below I' co by reversing the base 
current from its normal negative direction to the positive direction. 

The current reversal can also be regarded from the point of view 
of the current network shown in Fig. 2. If we put I e = 0, the leakage 
current I co flows from base to collector. Normally, the current al e 


Fig. 2 — Transistor currents In terms of s, l 6 and l co . 

flows from emitter to collector. The current flowing from emitter to 
base is the difference between the currents flowing in the other two 
branches, namely 

Ie — ale = (1— ot)I e . 
The base current will therefore be positive, zero or negative depending 
on whether I co is greater than, equal to, or less than (1— a)I e . 

The sign of the base current is sometimes important, because of the 
need to reverse this current. The signs of the collector and emitter 
currents do not need to be distinguished for practical purposes. The 
collector and base currents are no longer shown as negative in the 
published data. 


In audio output stages, driver stages, and switching circuits, the tran- 
sistors are required to handle large changes of current. The current 
amplification factors for large signals are distinguished by adding a bar 
to a and a', thus a (alpha bar) and a' (alpha dash bar). 

An alternative symbol for a' is Iife , the subscript p denoting 
forward transfer and the subscript e showing that the emitter is the 
common electrode (grounded-emitter connection). Capital letters are 
used in the subscript to indicate that the currents under consideration 
are direct and can be read from the static characteristic curves. 5' may 
also be represented by ft. 

Page 48 


The alternative symbol for a is Iifb • 

The large-signal current amplification factor is defined with particular 
reference to switching applications. The transistor is considered to be 
off when the emitter current is zero. In this condition, the leakage 
current or cut-off current I co continues to flow as a positive base current 
from base to collector. Suppose now that the transistor is switched on 
to some collector current I c . The change in collector current is 
numerically equal to I c — I C o • 

The base current lb in the on condition flows in the normal negative 
direction, out of the transistor. The change in base current is numeri- 


I b positive I b negative 

Fig. 3 — Relation of transfer characteristic in grounded emitter to transistor 


cally equal to lb + Ico . Hence the large-signal current amplification 
factor in grounded-emitter is 

., = Ale 

" Alb 

Ic — Ico 


where A represents a large change. This definition is illustrated 
graphically by Fig. 3. 

It is not usual to work in terms of a, but this quantity may be defined 


_ AI C Ic — Ico 

a ~Ai;- I e 

Ic — Ico 

~ Ic+Ib ' 

where the transistor is off, as before, when the emitter current is zero. 

Page 49 


RELATION OF l co TO l co 

When the base is open circuit and lb = 0, we have 

»c = le — A co 

and, from Fig. 2, 

(1— a)I e = I co . 

i' _ ^ co 

,co ~I^i 

= (l + a')Ico 
~ a'I co . 
This relationship is only true provided the value of a' inserted in the 

I e (positive) &'(I co +I b ) I c inegativ«) 

|lb(nnoy be + or-) 

Fig. 4 — Transistor currents in terms of 5', | b and l' co 

equation is correct for very low collector currents. This value is con- 
siderably lower than at normal operating currents. 

Fig. 1 is not to scale because, in order to make the stippled areas 
of appreciable size, I co had to be represented as a much larger fraction 
of l'co than it is in practice. Values are likewise not given on the scales 
of Fig. 3, and this figure too is not to scale. 


From Fig. 2 it follows directly that, numerically, 

Ic = Ico+aIe. •••(2) 

This is the fundamental equation for the transistor, and is true in all 

configurations. Eq. 2 is the equation to the straight line of Fig. 1. 

Sometimes it is more convenient to express the collector current 

in terms of I' co , a' and lb ; thus, numerically: 

Ic = Ico+a'(Ico+Ib) 

= I'co+a'Ib. • • .(2a) 

This equation corresponds to the diagrams shown in Figs. 3 and 4. 

Page 50 



The collector leakage currents I c0 and I' co , the current amplification 
factors a and a', and the base-emitter voltage Vbe are all temperature 
dependent to a greater or lesser extent. 

Temperature dependence is greatest for the leakage currents. For a 
rise in junction temperature from 25°C to 45°C, I co increases by a 

Tamb = 25°C 

T am bs45°C 








/ ? 








+ 10>lA 





100 -150 -200 


Fig. 5 — Effect of temperature on output characteristic in grounded emitter 
Fig. 6— Effect of temperature on input characteristic in grounded emitter 

factor of five, and I' c0 by a factor of eight, still considering the OC71. 

a increases by about 0-5 % and a' by about 1 % per degree Centigrade. 

Vbe decreases by roughly 2mV per degree Centigrade. This rate of 
change is almost constant for all transistors (germanium and silicon). 

The effect of temperature on the output and input characteristics is 
shown in Figs. 5 and 6. 

In Fig. 5, the full lines apply to an ambient temperature of 25 °C. The 
collector current is equal to the leakage current r co (at I b = 0) plus 
the current produced by biasing the transistor from lb = to, say, 
lb = — 30fxA. When the ambient temperature is increased from 25°C 
to 45°C, the increase in leakage current pushes the curves upwards 
relative to the I c scale. 

Consequently the curve at 45°C for I b = — 30£aA (broken line) lies 
above that at 25°C for I b = -60(jlA (full line). 

Page 51 


Fig. 6, the input characteristic, shows the possibility of reversing 
the base current. The effect of this reversal is shown on the output 
characteristic by the broken curve for lb = + 10(jiA. Reversing the 
base current allows the grounded-emitter transistor to operate at low 
collector currents, even at high temperatures where I' co reaches a high 


Transistor characteristics, like those of the thermionic valve, exhibit 
production spreads. Every endeavour is made to keep the spreads as 
small as possible. 

Spreads are of most importance when considering grounded-emitter 
operation. For the OC71, the collector leakage current I'co , which is 

V c = -4-5V 















/ y 


-50 -150 -250 -350 


Fig. 7 — Spread in base-emitter voltage of OC71 

nominally 150[iA, may attain an extreme upper limit of 325(xA. These 
figures apply at an ambient temperature of 25°C. At 45°C, the leakage 
current in the extreme case may be as much as 8x0-325 = 2 -5mA. 
A fairly wide spread also occurs in a' and a'. A spread of only a few 
percent in a introduces a much wider spread into the factor 1-a and 
hence into a', since 


A nominal a' of 47 for the OC71 (at a collector current of 3mA) 
corresponds to a maximum of 75, the spread Aa'/a' therefore being 
(75— 47)/47 ~ 0-5. A similar spread applies to 5c'. 
The spread in base-emitter voltage for the OC71 is shown in Fig 7. 

Page 52 



In thermionic- valve circuits, the normal cathode resistor reduces the 
influence of valve spreads on the position of the working point. 
Although there are a number of ways of introducing d.c. feedback into 
transistor circuits, the best of these are a refinement from the idea of 
using an emitter resistor. For transistors, d.c. stabilisation is parti- 
cularly important, because temperature effects are capable of introduc- 
ing a much wider variation in collector current than that due to 
spreads alone. 

The transistor should be biased by a method which prevents excessive 
shift of the d.c. working point. Insufficient d.c. stabilisation can give 
rise to the following effects: 

(a) Wide spread in input and output impedances. 

(b) Risk of overloading ('bottoming') at high ambient temperatures. 

(c) Possibility of thermal runaway. (This effect is normally important 
only in high-voltage and/or high-power stages.) 

The biasing circuits which follow are described as grounded base, 
grounded emitter or grounded collector, but this description applies to 

Fig.1 — Grounded-base circuit 

the d.c. conditions only, and does not prevent the use of the circuit in a 
different a.c. configuration. Any electrode can be grounded to a.c. by 
means of a large capacitance, or open-circuited to a.c. by means of 
inductance, without affecting the d.c. conditions. 


In the grounded-base configuration (Fig. 1), the base is biased with 
constant emitter current, by making the biasing voltage V e e large in 
comparison with the transistor input voltage V e b . 

This is the most stable arrangement. It gives the least change in 

Page 53 


collector current when transistors are replaced, and the least increase 
in current with temperature. 

The two equations governing the performance are: 

Ic = Ico+ale 



Ie = 

R ( 

Strictly speaking, the large-signal value a should be inserted in the 
first equation, but in practice, with small-signal transistors, little error 
is likely to arise in substituting the zero-frequency value <x . 

Assuming constant emitter current I e , the spread in collector current 
I c will arise from the small spread in <x , which is normally less than 4 %. 
At higher temperatures, the increase in the leakage current I co may 
become apparent, depending on the relative magnitude of I c and I co • 
A transistor operating at a relatively high collector current will 
be less affected by temperature. 

In practice, it may not be possible to satisfy the condition in which 
the bias voltage V e e is very much greater than the input voltage V e b , 
and the spread in V e b and its change with temperature may influence I e . 


In the simple grounded-emitter circuit (Fig. 2), the emitter is common 
to the input and output circuits. The transistor is biased with constant 
base current, as the battery voltage V cc is large in comparison with the 
input voltage Vbe • 

This is the simplest arrangement in that it requires only one battery 
and one resistor. Unfortunately the d.c. stability is poor. There is a 

>F*b J LOAD 

Fig. 2 — Simple grounded-emitter circuit 

large change in collector current on replacing transistors and a large 
increase in collector current with temperature, and there may be ther- 
mal runaway in high-voltage or high-power stages. Unless the intended 
application is extremely uncritical of these effects, this method of 
biasing is generally unsuitable. 

Page 54 


The performance equations are: 

Ic = I'co+a'Ib 

Vcc-V be 

Ib = 


The disadvantages of this method of biasing are, firstly, the collector 
current must always lie above I'co , which may become prohibitively 
high at high temperatures. Secondly, there is a wide spread in a' over 
a batch of transistors, and a' itself increases with temperature. Since 
Vbe is normally small in comparison with V C c , the spread in Vbe and 
its change with temperature are usually of only secondary importance. 


A collector-base feedback resistor (Fig. 3) is the simplest method of 
including some d.c. stabilisation in a grounded-emitter circuit. The 

Fig. 3 — Grounded-emitter circuit with feedback resistor 

Fig. 4 — Bypassing of feedback resistor to prevent a.c. feedback 

base resistor Rb is returned to the collector end of R c , instead of to 
the battery. The value of Rb is roughly equal to V c /Ib . Any increase in 
collector current causes a drop in collector voltage, and hence reduces 
the current flowing through Rb to the base, so compensating partly 
for the original change. 

With RC coupling, the collector resistor itself supplies the d.c. feed- 
back, and no additional components are required. There will also be 
a.c. feedback, unless decoupling is used as shown in Fig. 4, where Rb 
is made up of two approximately equal resistances. 

This arrangement provides some d.c. stabilisation, but again the 
collector current must always lie above I' co . In the limiting case, where 
the increase in ambient temperature causes the stage to bottom, the 
voltage across Rb will be zero, the base current zero, and the collector 
current equal to I'co. 

Page 55 


If the collector current in the circuit tends to change by AI C , the 
d.c. feedback will reduce this change to KxAI c (K < 1), where 

K = 


1 + 



K is the 'factor of stability' for the circuit. 

A small value of K gives the best stability, and may be achieved by 
making R c large or Rb small, the former implying a high battery 
voltage, and the latter a high collector current. 

By rearranging the circuit, it may be used for grounded-collector 

Fig. 5 — Grounded-emitter circuit for d.c. arranged in grounded collector to a.c. 
Fig. 6 — Emitter-resistor and potential-divider circuit 

operation, as shown in Fig. 5. Exactly the same equations apply, 
except that R c must be replaced by R e . 

In practice this circuit gives a K of about 0-7 or 0-8, so the degree 
of stabilisation is not great. Nevertheless it may be sufficient for some 


In the emitter-resistor and potential-divider circuit (Fig. 6), the 
emitter resistor introduces negative d.c. feedback. The input voltage 
Vbe is determined by the emitter resistor, in conjunction with the 
potential divider R1-R2 connected across the battery. 

This arrangement is by far the most commonly used, as the designer 
has most control over the stabilisation, while only one battery is 

Any increase in emitter current causes a large voltage drop across 
the emitter resistor, and reduces the base-emitter voltage. The base 
current is reduced and, because of the exponential shape of the input 
characteristic, there is a large degree of compensation for the original 

Page 56 


The feedback depends on R e , a high value giving better stabilisa- 
tion. The feedback also depends on how constant the base potential 
can be maintained during changes in base current; low values of Rl 
and R2 improve the stabilisation. 

At one extreme, if R e were large and Rl and R2 small enough to 
give a base potential of effectively zero resistance, the circuit would be 
indistinguishable from the grounded-base arrangement, with its inher- 
ently good stability. At the other extreme, if R e were zero and Rl and 
R2 very large, the circuit would become equivalent to the simple 
grounded-emitter arrangement, with its very poor stability. 

The circuit is very flexible, and a wide range of stability can be 
obtained between these two extremes. The collector current is no 
longer limited to the region above I' C o , and can be reduced almost to 


In practice the relationship between the values of R e , Rl and R2 
will depend on the particular requirements of the circuit, but certain 



Fig. 7 — RC-coupled circuit in grounded emitter to a.c. 
Fig. 8 — RC-coupled circuit in grounded base to a.c. 
Fig. 9 — RC-coupled circuit in grounded collector to a.c. 

limitations exist. The maximum value of R e depends on how much of 
the battery voltage can be dropped across it. The minimum values of 
Rl and R2 are dictated by the current which R1-R2 can be allowed to 
bleed from the battery, and/or by the shunting of the incoming signal 
when RC coupling is used. 

The factor of stability for this circuit is 

K = 



a'(Re+r e ) 

Rb+r b b'+Re+r e 

where Rb is the effective base resistance formed by Rl and R2 in 
parallel, and is given by 


Rb = 


Page 57 


and rbb' and r e are internal resistances in series with base and emitter. 

K will be low, and there will be good stability, if R e is high and Rb 

A low value of R b gives a low value of K, but accentuates the effects 
of changes in Vbe if Re is also low. 

The emitter-resistor and potential-divider circuit is suitable for RC 
or transformer coupling using any of the three a.c. configurations (Figs. 
7 to 12). 

N.T.C. Thermistor 

The circuit may be modified by replacing R2 by an n.t.c. thermistor 
(resistor having a negative temperature coefficient). However, a parallel 
combination of a normal and an n.t.c. thermistor is more usual (Fig. 13). 
As the temperature rises, R2 falls. The base voltage Vbe decreases, 
offsetting the rise in collector current. Variations in a' from transistor 

Fig. 10 — Transformer-coupled circuit in grounded collector to a.c. 
Fig. 11 — Transformer-coupled circuit in grounded base to a.c. 

Fig. 12 — Transformer-coupled circuit in grounded emitter to a.c. 
Fig. 13 — N.T.C. thermistor in potential-divider bias circuit 

to transistor are taken up as before by feedback through the emitter 
resistor, whilst the n.t.c. thermistor copes with the change in r co with 
temperature. Better stabilisation is therefore obtained. 

Complete compensation over the whole temperature range cannot 
be provided, as the law of an n.t.c. thermistor is approximately linear, 

Page 58 


while that of I'co is exponential. The same collector current only occurs 
at two fixed temperatures. 

P.T.C Thermistor 

A positive-temperature-coefficient (p.t.c.) thermistor is sometimes used. 

The change in collector current with temperature is caused mainly by 
the temperature dependence of the leakage current and of the base- 
emitter voltage. Under certain conditions, and particularly in power- 
output stages and for silicon transistors, the change in Vbe may become 
the more important of the two. 

The change in V be with temperature can be compensated by using 
emitter resistors of copper or nickel or other pure metal, which have a 
small positive temperature coefficient, instead of conventional resistors 
made from a zero-temperature-coefficient alloy. 

Overcompensation for changes in Vbe helps to counteract changes 
in leakage current. 

The p.t.c. thermistor may be particularly important in power- 
transistor circuits with low-impedance bias supplies. 


This circuit (Fig. 14) may be regarded as a special case of the preceding 
one, a base supply voltage Vbb providing the bias voltage instead of 

Fig. 14 — Two-battery and emitter-resistor circuit 

the potential divider R1-R2. Alternatively it may be regarded as a 
special case of the grounded-base circuit, in which a resistor Rb has 
been inserted in the base lead. 

The collector current flows through V cc and the emitter current 
( = I c +Ib) flows through Vbb . The base current is small, and a tapped 
battery can be used, as both parts will require replacement at about 
the same time. 

The same equation for the factor of stability applies to this circuit 
as the preceding one. 

The circuit saves one resistor with RC coupling, and two with trans- 
Page 59 


former coupling. There is no potential divider to increase the current 

Since the voltage Vbb of the extra battery will often be higher than 
that required for biasing the base, the resulting higher value of R e will 

f T T 

Fig. 15 — Transformer-coupled circuit in grounded emitter to a.c. 
Fig. 16 — Transformer-coupled circuit in grounded collector to a.c. 

give better stabilisation. A higher combined battery voltage is implied, 
which to some extent offsets the saving of the current drain through 

The design of the circuit is comparatively simple, as the current is 
almost entirely determined by Vbb and R e and little affected by Rb. 

This circuit is particularly suited to transformer coupling. With the 
transistor in grounded emitter or grounded collector to a.c. (Figs. 15 

Fig. 17 — RC-coupled circuit in grounded emitter to a.c. 
Fig. 18 — RC-coupled circuit in grounded collector to a.c. 

and 16), Rb is only the resistance of the transformer winding, and 
excellent stabilisation is achieved. 

The circuit does not show to the same advantage for RC coupling, 
because Rb must then be several k£2's, to avoid shunting of the input. 
Figs. 17 and 18 show the RC-coupled circuits, with the transistor in 
grounded emitter and grounded collector to a.c. 

Page 60 



Thermal runaway is a condition in which the collector current con- 
tinues to rise until limited by some external means, or until the tran- 
sistor is destroyed. The prevention of this effect is one of the chief 
purposes of d.c. stabilisation. 

Thermal runaway is easily avoided in RC-coupled stages. 

For a tiansistor to attain a state of thermal equilibrium, the junction 
temperature must reach some steady temperature Tj above the ambient 
temperature T am b. These temperatures are related by the equation 

Tj = Tamb + Qptot 

where ptot is the collector dissipation plus base dissipation, and is the 
rise in junction temperature per unit collector dissipation. In most 
small-signal applications, the base dissipation may be neglected, and 
ptot — Pc- 

On switching on the power supply, the following sequence of events 
takes place. The collector dissipation p c makes the transistor start to 
warm up, thus initiating a rise in junction temperature ATj . Because 
of the temperature dependence of I' co , a' and Vbe , the collector 
current will rise above its nominal quiescent value by an amount AI C . 
In its turn, AI C will lead to an increase Ap c in collector dissipation. 
Finally, Ap c leads to a further increase in junction temperature, equal 
to GAp c . 

Positive thermal feedback exists in the system, the loop gain being 


6Ap c 


If the loop gain is greater than or equal to one, the transistor will be 
thermally unstable, and the collector current will 'run away'. 

A simple expression can be derived for the loop gain. The collector 
dissipation is given by 

Pc = IcVce 
and if R represents the total d.c. resistance in the collector and emitter 

pc = Ic(V cc -IcR). 

Pc+Apc = (Ic+AI c )Vcc-R(Ic+AI c ) 2 
and by subtraction, neglecting (AI C ) 2 , 

Ap c = AI c (Vcc-2I c R) = AI c (2Vce-V C c). 
Thermal runaway cannot take place provided the loop gain G is less 
than one, but as the final collector dissipation is 1/(1— G) times the 
initial switching-on dissipation, it is advisable to design for a G of 
about 0-5 for a limit transistor. 

Page 61 


Thus the necessary condition for thermal stability becomes 

G = 6.^2 (2V ce -Vcc) < 1 ~ 0-5. . . .(1) 


If is given in °C per milliwatt, dI c /dTj should be expressed in milli- 
amps per °C, V ce and V cc being in volts. If is in °C/W, dI c /dTj 
should be in amps per °C. 

Half-supply-voltage Principle 

From Eq. 1, it follows that if the voltage across the transistor is 
equal to or less than half the supply voltage, the factor (2V ce — V cc ) 
will be zero or negative. G will also be zero or negative. Such circuits 
are inherently thermally stable. This is the half-supply-voltage principle. 
Most RC-coupled stages satisfy the half-supply-voltage condition 
without modification. 


Tables 1 and 2, at the end of the chapter (p. 72), give recommended 
circuits for the OC71. These circuits are suitable for operation up to a 
maximum operating ambient temperature of 45 °C. 

Examples have not been given "in Table 1 for RC-coupled circuits 
operating at collector supply voltages of less than 4-5V, because of the 
reduction in gain which occurs at lower battery voltages. For special 
applications, such as hearing aids, the use of RC coupling at low battery 
voltages may still be desirable, however. 


The working point must remain sufficiently outside the knee of the 
output characteristic to allow the required swing under drive, otherwise 
clipping will occur. Under extreme conditions, the increase in the direct 
collector current will 'bottom' the transistor. 

As the collector current rises, because of either a rise in temperature 
or transistor spreads, the working point moves up along a 'd.c. load 
line' (Fig. 19). This line is drawn across the output characteristics to 
pass through the collector supply voltage V cc , with a slope equal to the 
total d.c. resistance R c +Re in the collector and emitter circuits. The 
working point (I c , V c ) is given by 

V c = V cc -Ic(Rc+Re). -(2) 

The transistor bottoms when all the available voltage is dropped 
across the d.c. resistance, that is, when the collector current rises to 
a value 


Page 62 

Ic== Vcc-V k nee > ... (3) 



where Vknee is the knee voltage. 

The knee voltage depends on the collector current. When designing 
amplifier stages for the typical small-signal germanium and silicon 
transistors, it may be assumed that Vknee will not exceed 0-2V. 

It follows from Eq. 3 that stages in which V cc is increased or R c +R e 
decreased will not clip until a higher collector current is reached, that 
is, such stages can be operated up to higher ambient temperatures. 


Wherever possible, circuits for the OC71 should be selected from those 
given in Tables 1 and 2. A design procedure will now be described 
which is suitable for special requirements, such as higher supply 
voltages, and for transistors other than the OC71. 

There are many ways of proceeding with the design, but this method 
has been chosen as being sufficiently rigorous and simple. Circuits 
designed along these lines will have if anything a larger margin of 

T am b s 25 C 

T am b = 45 C 



a0m . — 







\ ? 









+ 10mA 



Fig. 19 — Output characteristics with superimposed d.c. load line 

safety than is really necessary, at the expense of some loss of gain and 
slightly increased battery drain. For 'one-off' quantities, however, the 
ultimate in performance is usually not essential. The circuits may be 
incorporated in permanently installed equipment, and need not be 
restricted to 'breadboards' or 'lash ups'. 

The method is based on three rules of thumb: 

(a) The circuit should be designed for a collector current of 1mA, 
unless the signal level requires a higher current. 

Page 63 


(b) About \ to IV should be dropped across the emitter resistor R e . 
For a collector current of 1mA, the emitter resistor will be in the 
region of 470Q to lkQ. 

(c) The external base resistance Rb normally should be chosen to be 
of the order of 10 X R e . The circuit will then generally cope with 
junction temperatures of up to 45 to 50°C. 

More flexibility is possible in the circuit design than these rules of 
thumb might suggest. The collector current may be increased above 
1mA if the signal swing demands it. Collector currents below 1mA 
should be avoided as far as possible, because better stabilisation 

Fig. 20 — Equivalent circuit for one-battery and two-battery circuits 

becomes necessary. The stabilisation may also have to be improved 
if the transistor will be expected to operate at high ambient tempera- 
tures, and/or if the supply voltage is low. The improvement is effected 
by increasing the voltage dropped across the resistance in the emitter, 
and by decreasing the ratio of Rb/Re from 10 to perhaps 5 or less. If 
Rb is very low, then it may be necessary to include rbb' in Rb , because 
Tbb' will no longer be negligible in comparison with Rb. 

The one-battery and two-battery circuits can be analysed by means 
of the same equivalent circuit (Fig. 20). From the equivalent circuit, 
it follows that 

Vbb = IbRb + Vbe + IeRe. -(4) 

Since the emitter and collector currents are approximately equal, and 
since the base current is approximately given by I c /«\ 

V b b 


+ ^|I C +Vbc 


For small-signal transistors, at collector currents of a few milliamps, 
a' may be taken as being equal to <x' . 

In Fig. 21, Vbb is plotted against Rb with R e as parameter for a 

Page 64 


collector current of 1mA, the transistor being the OC71, for which 
a' ~ 40 and V be ^ 0-1V. 

The value of Rb is chosen with the aid of Fig. 21 or Eq. 4a. In the 
two-battery circuit, Vbb is fixed in advance, and there is some restriction 
on the choice of component values. In the one-battery circuit, the base 
can be regarded as being fed from an artificial tap on the battery, the 








Fig. 21 — Base supply voltage related to external base resistance and to resist- 
ance in emitter circuit. 

proportion of the voltage tapped off being completely under the 
control of the designer. 

From the values of R b , Re and a', the factor of stability K can 
be calculated, using 

(Rb+Re) 1 

K = 

(Rb+Re)+a'R e 


a'R € 



If the factor (Rb+R e ) is particularly small, rbb' must be added to Rb. 
K also can be found from a graph, such as that given in Fig. 22 for 
an a' of 75. 

The changes in the collector current caused by (a) the spread in a' , 
(b) the increase in leakage current with temperature, and (c) the 
decrease in Vbe with temperature and the spread in Vbe , are additive 
and are given by: 

Page 65 




AI C( 2) ~ K(l+a')AI C0 , 
at Ka ' 

AIc(3) = -r^+r; 


• AV t 





The total change in collector current 

Ale = AIc(i) + AIc(2)+AI C (3) 

gives the maximum collector current as 

I c max = I c nom+AI c . 

Eq. 6 allows for the effect of inserting a transistor having the maxi- 
mum a' in a circuit designed for the nominal a. K is calculated for 


i*bb' *■ r e neglected 



l 3 / 




*" n.p- 








Fig. 22 — Evaluation of K for an a' of 75 

the maximum a'; Aa' is the difference between the maximum and 
nominal a'; and the nominal value of a' is inserted in the denominator. 
Usually Aa'/a' ~ 0-5. 

Eq. 6 may also be used for calculating the effect of inserting a 

Page 66 


transistor having the minimum a'. K should then be calculated for the 
minimum value of a'. Aa' will now be negative and AI C (i) will be 

Eq. 7 primarily allows for the increase in leakage current up to the 
maximum junction temperature occurring in the circuit. The increase 
can be found by means of a curve such as that shown in Fig. 23, which 
is for the OC71. The calculation should be made for a transistor 
having the maximum I co and — although this combination is rather 
unlikely in practice — the maximum a'. For a transistor operating under 



25 30 35 40 45 50 55 

Fig. 23 — Collector leakage current in grounded base, l co . at a given junction 
temperature divided by Its value at 25°C. 

true small-signal conditions, the maximum junction temperature may 
be taken as being approximately equal to the maximum ambient 

With reference to Eq. 8, Vbe for both silicon and germanium tran- 
sistors decreases by roughly 2mV for every °C rise in temperature. The 
sign of AI C (3) is positive for a decrease in Vbe and negative for an increase 
in Vbe- Eq. 8 should be used for the one-battery case, but 8(a) is 
sufficiently exact for two-battery circuits. 

The minimum acceptable collector-emitter voltage is 

„ . „ T , . nominal V c 
V c min = (Vknee+e c ) X 


' end-of-life V cc 

where e c is the peak signal required. No end-of-life correction is 
needed for a mercury cell, which has a constant-voltage characteristic. 
Nor is this correction required if the stage drives another whose 
handling capacity is also reduced as V cc falls. 

The resistance required in the collector is found from: 

Vcc— Vcmin 




I max 

If the value of R c is unacceptably low, then the circuit must be re- 
designed for a higher battery voltage, or for lower values of Rb and K. 

Page 67 


The junction temperature at the maximum collector current is : 

max.Tj = T a mbmax+e(I c max)(V c min) ...(13) 

where 6 is expressed in °C/mW and is 0-4 for the OC70, OC71 and 
OC75. If the maximum junction temperature exceeds the maximum 
ambient temperature by more than 2 to 3°C, the circuit must be re- 
designed using the true junction temperature when calculating the 
change in leakage current. 

For the one-battery circuit, the values of the resistors in the potential 
divider are found from : 

R 1= %*» ...(14) 


and R a = Voe *» • -(15) 

Vcc — Vbb 

If Rl and R2 are fairly similar in value, the tolerances should be 
±5%. Otherwise ±10% is sufficient. 

In the following examples, no allowance has been made for the 
spread in Vbe , but only for the change with temperature. There is 
little error, however. The nearest lower preferred value is chosen for 
R2and R c . 

Example 1 

Tapped-battery RC-coupled small-signal circuit for OC71: V cc = — 9V, 
Fbb = —1-5V, max. operating T^mb = 45° C. 
Choose Ic = 1mA and R e = 1 -2kQ. 
From Fig. 21, R b = lOkQ. 
Maximum a' ~ nominal a'-}- 50% 

= 40+20 = 60. 
The value of K for an a' of 60 is, from Eq. 5, 

(Rb+Re) _ 11-2 _ 

(Rb+Re)+oc'R e 11 -2+60x1-2 
From Eq. 6, 

AI C (i) = K^tVc = 0-13 x 0-5 X 1 -0 = 0-07mA. 

From Eq. 7 and Fig. 23, where I co max = 13[xA, 

AI C(2 ) ~ K(l+a')AI co = 0-13 X 61 X (4 X 13 X 10" 3 ) = 0-4mA. 
From Eq. 8(a), 

AI C( 3) x ATambX2mV/R e = (45 -25) X 2/1 -2 = 0-03mA. 
From Eqs. 9 and 10, 

Icmax = l-0+0-07+0-4+0'03 = l-5mA. 

Page 68 


From Eq. 11, for no end-of-life correction, 

Vcmin = Vknee+e c = 0-2+0-1 = 0-3V. 
From Eq. 12, 

Re = V ee-Vcmin _ = ^^ ^ 3 . 9kQ±10 o /o . 
I c max 1-5 

The voltage dropped across the d.c. resistance is 

V R = I C (R C +Re) =1-0(3-9+1-2) = 5V, 

and as this is greater than 4-5V( = |V C c) the circuit is necessarily 
thermally stable. 

Example 2 

One-battery RC-coupled small-signal circuit for OC201: V cc = —12V y 
max. operating r am b = 100°C. 

The OC201 is a silicon transistor having an I co of only lOfxA at 
Tj = 100°C, and of only lOOmjxA at 25°C. 

8 = 0-5°C/mW. 

Stabilisation is now primarily for the spread in a' and change in Vbe- 

Choose I c = 1mA and R e = 470ft. 

Choose Rb = 5kft. 

At 1mA, the nominal a' = 30 and the maximum a' = 80. 

\r = 19 


Of) -3Q 

AIc(i) = 0-12x-^-^xl-0 = 0-2mA. 

AI C (2) = 0-12x81 x(10xlO- 3 ) = 0-lmA. 

Al c(3) = °' 1 ^ 81 -(100-25)x2xl0- 3 = 0-26mA. 

Icmax = 1-0+0-2+0- 1+0-26 = l-56mA. 

V c min = 0-2+0-1 = 0-3V. 


Re = — t-2 0-5 ~ 5-6kQ±10%. 


V R = (5-6+0-5)1-0 = 6-1V > IVcc 

For a one-battery circuit, V b b , Rl and R2 have to be calculated 
using Eqs. 4(a), 14 and 15. 

V bb ££ (Re+^Ic+Vbe 
Page 69 


1-0+0-1 = 0-77V. 


Rl = Y^> = l l^A ~ 68kft±10%. 

Example 3 

One-battery RC-coupled large-signal circuit for OC71: V cc = —14 V, 

max. operating TWb = 45°C. 

Choose Ic = 3mA. 

In view of the high supply voltage, R e may be chosen as high as 470Q. 

R b may be chosen as 5kQ ( ~ 10xR e ). 

At 3mA, the maximum a' is 75. Assume max.Tj = 47°C. 

From Fig. 22, 

K = 0-14. 

AIc(i) = 0-14x0-5x3 = 0-21mA. 

AI C( 2) =0-14x76x(5xl3xl0- 3 ) - 0-69mA. 

AIc(3) = °' 14x75 .(47-25)x2xl0- 3 - 0-08mA. 

I c max = 3+0-21+0-69+0-08 = 4mA. 
V c min = 0-2+0-5 = 0-7V (for 1-0V peak-to-peak) 

r c = 1 izP_^_o-47 - 2-7kQ±10%. 

The maximum junction temperature, from Eq. 13, is 
max. Tj = T a mbmax+6(I c max)(V c min) 
= 45+0-4x4x0-7 = 46°C. 
The calculation does not need to be repeated, since a max. Tj of 47°C 
was assumed. 

V B = (2-7+0-5)3 = 9-6V > |V CC . 
As the design is satisfactory in the above respects, V b b , Rl and 
R2 may be calculated. 

Vbb ^ (o-5+J-Wo+O-l =* 1-9V. 

~- Kl>) 

Rl = t^t ~ 33kO±10 °/- Rz ~ i4^ii - 5 - 6kn ± 10 %- 

Page 70 


Example 4 

One-battery RC-coupled large-signal circuit for OC75: V C c = —18 V, 

max. operating 7 , a mb= 45° C. 

The OC75 has a maximum a' of 130 at 3mA and a maximum I c0 of 

14[xA at 25°C. 

Choose Ic = 3mA. 

R e may be chosen as 680O, as there is plenty of supply voltage available. 

Choose R b = 7kQ ( ~ 10 xR e ). 

Assume max. Tj = 47°C. 

K = 7-7+130X0-7 = °'° 8 - 

AIc(i) = 0-08x0-5x3 = 0-12mA. 

AI C(2) = 0-08 x 130 x (5 X 14 x 10~ 3 ) = 0-73mA. 

AI C (3) - '°^ 130 -(47-25)x2xl0- 3 = 0-06mA. 

I c max = 3+0-12+0-73+0-06 = 3-9mA. 
V c min = 0-2+0-8 = 1-0V (for 1-6V peak-to-peak) 

io 1 

Re = -j^ 0-7 = 3-3kO±10%. 

max. Tj = 45+0-4(3-9)(l -0) = 46-6°C. 
V R = 3(3-3+0-7) = 12V > £Vcc 

V bb = /o-7+^Wo+O-l = 2-36V. 

R2= ira6^ 6 - 8kn±10% - 

Page 71 



max. operating T am b = 45°C 























































12 1-5 1-0 56 10 4-7 

Transformer Coupling 

























*Resistance of primary winding of output transformer 


V cc = — 6V, Vbb = — 1-5V, max. operating T am b — 45°C 



Re Rl) 

(kQ) (kQ) 
RC Coupling 




2-7 10 
2-7 6-8 
1-2 10 

Transformer Coupling 









•Resistance of primary winding of output transformer 

Page 72 



The major processes occurring in a junction transistor have already 
been outlined in Chapter 1. These processes will now be represented 
by electrical circuit elements, which will be assembled to form a com- 
plete equivalent circuit. 

By making some approximations, the complete equivalent circuit 
can be simplified into a number of more familiar forms. The relation- 
Base width. — H [-'—Depletion loyer 



Emitter 'junction b A Collector junction 

-T 1 T" + 

Fig. 1 — Theoretical model of junction transistor 

ship between the most important of these circuits is shown, and some 
of the approximations made in their derivation are pointed out. 

A family of six circuits is described, each member of which covers 
the full useful range of operating frequency. Three circuits will also 
be given which apply to restricted frequency ranges. 

From the values of not more than six transistor parameters, any of 
the equivalent circuits can be written out in full. For some of the 
circuits, fewer than six parameters need be known. 

For reference, the six basic parameters are listed here in the order 
in which they will be introduced: r e , a , [x , rb»v , c<iep and c e . 

Three equivalent circuits are used as a basis for drawing up the 
published data. They are the low-frequency T circuit, the hybrid tc 
grounded-emitter circuit, and the complete grounded-base T circuit. 

The approximate equivalent circuits described in this chapter apply 
to alloy-junction transistors only. 


The transistor may be represented by the theoretical model of Fig. 1. 

Emitter Action: Carrier Injection 

As all the current carriers leaving the emitter flow into the base, no 

Page 73 


components are required to represent carrier injection. The emitter 
is therefore shown connected directly (Fig. 2) to the components 
which represent the diffusion of carriers through the base. 

The carriers leaving the emitter will be holes in a p-n-p transistor, 
or electrons in an n-p-n one. In what follows a p-n-p transistor will 
be assumed, for the sake of definiteness. 

Diffusion through Base 

Holes (in a p-n-p transistor) cross the emitter junction and diffuse 
within the field-free region of the base, until they reach the collector 

Fig. 2 — Representation of diffusion and collector action 

depletion layer. This process is represented by the symmetrical RC 
transmission line shown in Fig. 2. 

The holes exist long enough in the base for a proportion of them to 
recombine with current carriers of opposite sign. Recombination in 
the base region is represented by the distributed shunt resistance along 
the transmission line. 

The length of the transmission line corresponds to the width of the 
base. Recombination is less in a narrower base. 

Collector Action 

All the holes reaching the collector depletion layer, during their random 
movement of diffusion in the base, will be swept out of the base into 
the collector by the field existing across the depletion layer. This process 
corresponds to a short-circuit on the end of the transmission line, as 
shown in Fig. 2. 

The flow of current carriers must appear finally as a current at the 
collector terminal flowing out of the reverse-biased collector junction. 
Since in short-circuiting the line the effect of this current has been 
removed from the equivalent circuit, an infinite-impedance current 
generator is connected between the collector and base terminals in 
Fig. 2, to restore it. The current flowing from the current generator 
equals that flowing in the short-circuited end of the transmission line. 

Fig. 2, which still does not give a complete representation, may 

Page 74 


be compared with the first low-frequency equivalent circuit proposed 
for the transistor (Fig. 3). 

Emitter Resistance r e 

The input resistance of the short-circuited transmission line is known 
as the emitter resistance r e , and is given by the formula 


r e 


where T is the junction temperature measured in °K and I e is the 

Fig. 3 — First low-frequency equivalent circuit 

emitter current in milliamps. An approximate but more useful expres- 
sion, applicable at room temperatures (18 to 20°C), is 

r e — — ohms. 

I e may be 1mA and then r e is 25£2. r e is proportional to 1/I e and to 

T (°K). The temperature in degrees Kelvin is °C+273. 

Current Amplification Factor 

The forward current transfer in grounded base with the output short- 

Fig. A — Complete equivalent circuit with transmission line 

circuited to a.c. is the current amplification factor a = ia/ie . The value 
of a at low frequencies is denoted by a and is usually about 0-98. 

A number of components will now be added to Fig. 2 in order to 
derive the complete equivalent circuit shown in Fig. 4. 

Page 75 


Width of Collector Depletion Layer: Feedback Factor fx 

The width of the space-charge layer or depletion layer shown in Fig. 1 
depends on the collector voltage (Early effect). If the collector voltage 
is increased, the width of the depletion layer increases, and so the 
effective width of the base decreases. Therefore the base width, and 
the length of the transmission line, vary with collector voltage. 

The modulation of the base width by the collector voltage is repre- 
sented by a zero-impedance voltage generator in the short-circuited 
end of the transmission line (Fig. 4). The voltage of the generator is a 
fraction y. of the voltage V' c across the collector junction. 

Base mote riot 

Fig. 5 — Diagram of cross-section of practical junction transistor 

The significance of \x. is that it represents one form of feedback 
within the transistor. 

If [i is put equal to zero, the internal feedback at low frequencies 
becomes zero, and the equivalent circuit becomes that much nearer 
to the simple one shown in Fig. 3. 

\i is proportional to , and is independent of frequency. 

Capacitance of Collector Depletion Layer c de p 

The collector depletion layer may be regarded as a parallel-plate 
capacitor Cdep , which is included in Fig. 4. A typical value for Cdep is 
lOpF. This capacitance causes feedback within the transistor at high 

Cdep is proportional to , . 

V »c 

Ohmic Base Resistance r bb - 

In a practical transistor, shown diagrammatically in Fig. 5, it is neces- 
sary to take into account the resistance of the base material between 
the active part of the transistor (between emitter and collector) and 
the base connecting wire. The external base connection ideally would 

Page 76 


be made to an imaginary point b', in the active part of the base material. 
The resistance between b' and the actual base connection b is the 
internal base resistance rbb' included in Fig. 4. rbb' depends on the 
construction of the transistor, and a typical value (e.g. for the OC45) 
is 75D. 

The complete equivalent circuit has now been assembled and all 
the components shown in Fig. 4 have been identified. This circuit is 


1 Ml 


;C2-ce(-^hJV & , 2 

Fig. 6 — Simplified equivalent circuit with collapsed transmission line 

too complicated for most practical purposes. It will now be simplified 
and transformed into the more usual arrangements. The T equivalent 
circuits will be considered first, then the n circuits. 


First an approximation will be made to the transmission line, and then 
part of the circuit will be converted into an alternative network having 
identical properties. 

In Fig. 6, the circuit of Fig. 4 has been simplified by lumping the 
total resistance and capacitance shunted across the transmission line 
into two equal parts, and connecting half the total at the start of the 
line and half at the finish. Thus Ri = R 2 and Q = C 2 . Pictorially 
this process is equivalent to collapsing or telescoping the line, R3 
being the total series resistance of the line. 

Rl, R2 and R3 are all simply related to r e by factors of a Q and 
1— a Q , as shown in Fig. 6. Rl and R2 are very much greater than 
R3, since 

Rl _ oco _ , 

R 3 ~ 


a Q 

Emitter Capacitance c e 

CI in Fig. 6 is known as the emitter capacitance c e . C2 also equals c e . 

All the six basic parameters listed at the beginning of the chapter 
have now been introduced. 

Page 77 


c e is proportional to I e and to 1/Tf a , where T is in °K and f a is 
the a cut-off frequency. 

Network Conversion 

Interchanging R2 and C2 in Fig. 6 does not affect the electrical pro- 
perties of the circuit or in any way disturb the electrical symmetry. 
The central part of the network shown in Fig. 7 may be extracted and 
considered separately. 

The central part of Fig. 7 can be converted into a different network, 
shown in Fig. 8, without in any way altering its electrical character- 

Cl-c e 

Fig. 7 — Previous figure redrawn and divided into three 

istics. The characteristics of the whole network will therefore also be 

The proof that the central part of Fig. 7 is identical with Fig. 8 runs 
as follows. The performance of this type of network may be described 
completely by means of four parameters. If the four parameters 
describing one network are identical with those describing another 
network, then all the characteristics of the two networks are identical. 
It is convenient to choose the input and output impedances for open- 
and short-circuit terminations as the four parameters. Then the two 
networks are identical since for both: 

Input impedance with output terminals short-circuited = 
Input impedance with output terminals open-circuited = R2 
Output impedance with input terminals short-circuited = 
Output impedance with input terminals open-circuited = R2/H- 
A typical value for R2 is 1250Q and (x could equal 1/2000, making 
the value of R2/V in Figs. 8 and 9 equal to 2-5MQ. 

C2 in Fig. 6 can be transferred to the right-hand side of the generator 
by the same method as used for R2. A similar change in impedance 
is made by multiplying the capacitance by [i, to give (i.c e ( = (XC2) 
in Fig. 9. 

Page 78 


Collector Capacitance and Collector Resistance 

The combined capacitance of (xc e and Cdep in parallel is known as the 


Fig. 8 — Alternative form of centre network 

collector capacitance c c . For most purposes jj.c e is negligibly small in 
comparison with c<iep , and c c m Cdep • 

Fig. 9 may now be simplified to Fig. 10. With V' c = 0, the input 
resistance between b' and e is equal to Rl and R3 in parallel, that is, 
r e . When V' c is not zero, the voltage yiV'c appears through R3 at the 

i, R3 i 2 
-5 f-VWv >- 

©fpVi i 2 K 

! c dep 


>r bb' 

Fig. 9 — R2 and C2 transferred to collector 

Fig. 10 — Two-generator grounded-base T equivalent circuit 

input terminals. Because R3 = r e /a s r e , the (xV' c generator may 
be placed in series with r e as shown in Fig. 10. 

The effect of Rl on the output resistance has now been removed, 
so it must be restored by connecting an additional resistance across 
b' and c, as was previously done for R2. Because R3 is very much 
smaller than Rl, the effect of Rl on the output resistance is approxi- 
mately the same as that calculated for R2. Since two equal resistances 

Page 79 


in parallel give a combined resistance equal to half either one of them, 
the total resistance between b' and c in Fig. 10 becomes half that in 


r bb'< 



Fig. 11 — Usual form of two-generator grounded-base T equivalent circuit 

Fig. 9. This resistance is known as the collector resistance r c (= R2/2[a). 

a is equal to i2/i e , and at low frequencies the effect of c e is negligible, 
so that ii = i e and a = i2/ii . The current generator in Fig. 10 is 
therefore labelled a ii. 

The complete grounded-base T equivalent circuit shown in Fig. 10 
is redrawn in its more usual form in Fig. 1 1 . 

A number of equivalent circuits are related directly to the grounded- 
base T circuit, and these will now be described, but without giving the 
full derivation. 


The double-base-resistance equivalent circuit of Fig. 13 is equivalent 

-Si-ww— (~K 



i — VWv — * i f — ° c 



> r bb' 

Fig. 12 — Redrawn version of grounded-base T equivalent circuit divided into two 

electrically to the grounded-base circuit of Fig. 11. To show the 
relationship between the two, Fig. 11 has been redrawn in Fig. 12. 

The double-base-resistance circuit (Fig. 13) is obtained by substitut- 
ing the network within the box for the corresponding network in 

Page 80 


Fig. 12. The conditions for the two networks to be identical are found, 
as before, by equating the values of four parameters for each network. 
These conditions are included in Fig. 13. 

The double-base-resistance circuit contains only one generator 
(the current generator). The effect of the form of internal feedback, 



r" b -pr e 

rJ = r e -(l-o< ) pr c 

' r bb' 




Fig. 13 — Double-base-resistance grounded-base equivalent circuit 

previously accounted for by the jxV' c generator in the emitter circuit, 
is now taken care of by the factor (x, which appears in each of the 
expressions for r"b , r" e , r" c and <x" . 


If the capacitors are omitted from the double-base-resistance equivalent 

Fig. 14 — Conventional low-frequency T equivalent circuit for grounded base 

circuit (Fig. 13), the result is the grounded-base low-frequency T 
circuit shown in Fig. 14. 

In Fig. 15 the grounded-base circuit has been redrawn with its emitter 
grounded. This is the grounded-emitter low-frequency T circuit. The 
quantities r* e , r" c and (r"b+rbb') are now represented by capital 

Page 81 


letters as R e , R c and Rb . Often they are quoted simply with small 
letters as r e , r c and rb , but this practice can be confusing since r" e 
equals r e /2. 

Rb (or rb) is rbb'+[*r c and R c ^ r c . 

This equivalent circuit is given in the published data for the OC70, 
OC71 and OC75 low-frequency transistors. 

An alternative form of the circuit with the current generator expressed 
in terms of a' and ib is shown in Fig. 16. To keep the output impedance 

Fig. 15 — Low-frequency T equivalent circuit for grounded emitter 

the same, R c has been replaced by R c (l— a ), which is approximately 
equal to R c /a' . 


In the equivalent circuit of Fig. 11, the voltage V' c that can be 
developed across the collector circuit at high frequencies is usually 
severely limited by the low reactance of c c , and so the voltage [iV'c 


bo > AAAA- 

» \AA/V— — * oe 

R c (!-o<o) 
sR e /.T 

Fig. 16 — Alternative low-frequency T equivalent circuit for grounded emitter 
Fig. 17 — Simplified high-frequency T equivalent circuit 

in the emitter circuit may be neglected. Similarly, since c c is a low 
reactance, little current flows through r c , and this also may be neglected. 
The high-frequency circuit therefore reduces to that of Fig. 17. 


The grounded-base T circuit shown in Fig. 10 can be transformed into 

Page 82 


the 7r circuit shown in Fig. 18. The circuit values have been calculated 
by the same methods as for the other circuits, by comparing alternative 
portions of the circuits and equating parameters. The resistance in 
parallel with c c is equal to 2xr c ( = R.2/^). 
g m is the mutual conductance of the transistor, and its value is 


gm=— • 


The hybrid n grounded-emitter circuit is perhaps the most generally 


Fig. 18 — Complete grounded-base re equivalent circuit 

useful, and will therefore be dealt with in slightly more detail than 
some of the preceding circuits. 

Derivation from Grounded-base tc Circuit 

The hybrid tt grounded-emitter equivalent circuit can be derived from 
the grounded-base n circuit. In Fig. 19 the circuit of Fig. 18 is redrawn 

Fig. 19 — Hybrid n grounded-emitter equivalent circuit (rearrangement of 

previous figure). 

with the emitter connection common to the input and output; otherwise 
the two circuits are identical. 

Fig. 19 may be transformed further into Fig. 20 by replacing the 
current generator by two identical generators in series, with their 
junction connected to the emitter. 

Page 83 


The current leaving the point b' and flowing into the generator is 
the same in both circuits. Similarly the current entering the point c is 
the same. The current entering the common line e-e from the left-hand 
generator is extracted by the right-hand generator, and since both 
generators are by definition infinite impedances, the conditions at e-e 
are unchanged by connecting the generators. Consequently Figs. 19 
and 20 are identical. 

In Fig. 20 the left-hand generator is a function of the voltage across 
its terminals. Therefore it can be replaced by a resistance, which is 

Fig. 20 — Hybrid n grounded-emitter equivalent circuit with duplicate current 


given a negative sign because of the direction of current flow through 
the generator. The resistance can be shown to be — 1/gm , or — r e /a 
ohms. This resistance is in parallel with r e , so the total resistance 
rb'e between the points b' and e can be worked out from 
1 1 a 1— a Q 


r e r e 


r b'e 

Complete Hybrid 7t Grounded-emitter Circuit 

The hybrid w equivalent circuit, which is used to describe the pro- 
perties of high-frequency transistors in published data, is shown in 
Fig. 21, with its values stated in terms of the six basic parameters. In 
Fig. 22, the hybrid n circuit is shown with the circuit elements labelled 
with their usual symbols. 


With the exception of the low-frequency and high-frequency circuits 
given in Figs. 3, 14, 15, 16 and 17, all the circuits give useful results 
up to the a cut-off frequency. Errors are greater, however, at the higher 
frequencies. The circuits are intended for small-signal operation under 
normal d.c. operating conditions, for example, as a class A amplifier. 
No account has been taken of surface effects, collector-junction 

Pace 84 


leakage, emitter efficiency and so on. In practice these factors may 
entail some modification to the equivalent-circuit values. 


The first complete equivalent circuit (Fig. 4), which includes a trans- 
mission line, is the most accurate. It is closely related to the physical 
operation of the transistor. This circuit and its first approximate 
version (Fig. 6) are most useful for studying the transistor in its own 
right, as distinct from using it for particular applications. 

For application work any of the following four conventional circuits 
may be used: 

two-generator grounded-base T circuit (Figs. 10 and 1 1) ; 

double-base-resistance grounded-base circuit (Fig. 13); 

grounded-base tz circuit (Fig. 18); and 

hybrid n grounded-emitter circuit (Fig. 21). 
All these circuits convey the same information with equal accuracy. 

For grounded-base operation, the choice between the three grounded- 
base equivalent circuits will depend on the particular application, and 
to some extent on the personal preference of the circuit designer. 

For grounded-emitter operation, the hybrid n circuit is probably 
the most generally useful, because the feedback within the transistor 

Fig. 21- 

-Values of hybrid ji grounded-emitter circuit related to grounded- 
base circuit of Fig. 10. 




9m v b'e 

Fig. 22 — Usual symbols for hybrid n grounded-emitter equivalent circuit 

is expressed in its simplest form, so allowing the performance of the 
transistor to be estimated from an inspection of the circuit. 

The various approximate versions of these circuits can also be 
extremely useful in circuit design, provided their limitations are assessed 

Page 85 





-HI — 

c .!OSpF 


r b-»"c 
rJ-r e -(l-o< ) |Jr c 

i"J -r c (i-»i)ir c 


do lb 

bo > VWV 

R b I R, /tt '.r" (l-a )*25kn 

"b+rbb-TOOn J e ' a ° c ° 

')>R e =r e /2 
. I25fi. 


2r c -2 5Mfl 

(•) r^r 

Fig. 23 — Representative values inserted on five equivalent circuits 
Page 86 


for the particular application. For example, in the design of a low- 
frequency amplifier having a gain of only a few times, the first approxi- 
mate equivalent circuit shown in Fig. 3 would normally be quite ade- 
quate. At the other extreme, where no simplification is possible, such 
as in the design of some moderately high-frequency amplifiers, the 
equivalent circuit merely forms the basis for the calculation of a less 
complex but frequency-dependent network, which represents the 
transistor at one frequency only. 


As a numerical example, the following values will be assumed for a 
hypothetical transistor: 

a = 0-98 (a' ~ 50) 

r e = 25Q. 

y. = 1/2000 
rbb' = 75Q 

Cdep = 10pF 

and c e = lOOOpF. 

These values have been chosen to give round numbers, and while they 
do not apply to any particular type, they are sufficiently representative 
of an r.f. transistor. 

Five equivalent circuits will now be written out using the above 
values (Fig. 23). 

(a) The two-generator grounded-base T circuit is shown in Fig. 23(a). 
The values of r e , c e , \x, rbt>' and a can be written directly on to the 
circuit. The only quantities to calculate are c c and r c , thus: 

C c = Cdep+^Ce 

= 10-5pF; 

and r c 

2(x(l— <x ) 
= Jx25x2000x50 
= 1-25MG. 

(p) The double-base-resistance circuit is shown in Fig. 23(b). On to 
this circuit we can write immediately : 

c e = lOOOpF and r bb ' = 75Q. 
Page 87 


Also, c c has already been calculated as 10-5pF. The other quantities 
(using r c = 1-25MQ) are: 

r" e = r e /2 = 12-5Q; 
r" b = fxr c 

= l-25xl0 6 /2000 = 625Q; 
r"c = r c (l-(x) 

~ r c = 1-25MQ; 

„ a — fi. 

and a o = ~. - 


~ a = 0-98. 

(c) The low-frequency T circuit arranged for grounded emitter is shown 
in Fig. 23(c). Here the quantities are calculated as: 

Rb = r"b+rbb' 

= 625+75 = 7001i; 
R e = r e /2 = 12-50; 
a' = a Q /(l — a ) 

= 0-98/0-02 = 49; 

and Rc/a'o ca r" c (l — a ) 

= 1 -25 X 10 6 /50 = 25kQ. 

(d) The complete grounded-base tc circuit is shown in Fig. 23(d). The 
following values can be filled in straight away from the six basic 
parameters : 

c e = lOOOpF; r e =. 250; r bb » = 75Q; and a = 0-98. 
Also, c c = 10-5pF as calculated previously. Then 
r e /n = 25x2000 = 50kQ; 
gm = a /r e = 0-98/25 = 39mA/V; 

and ,/* , = 2r c = 2-5MQ. 

[a(1 — a ) 

(e) The hybrid n grounded-emitter circuit is shown in Fig. 23(e). The 
basic parameters on this circuit are: 

rbb' = 75Q; and c b - e = c e = lOOOpF. 

From previous calculations : 

Cb-c = c c = 10-5pF; r b 'c = 2r c = 2-5MQ; and g m = 39m A/ V. 

It only remains to calculate: 

r b -e = r^— = 25x50 = l-25kD; 
1— a 

and r ce = r e /n = 25 x 2000 = 50kQ. 

Page 88 


Introduction: the Idealised Transistor 

The complete equivalent circuit shown in Fig. 4 consists of 

(a) a transmission line and a current generator, and 

(b) Cdep , [>-, and rbb- . 

The second group of quantities may be classed loosely as parasitic 
elements, since they are in no way fundamental to the operation of the 
transistor. They will be neglected, therefore, in this section, and only 
an ideal transistor consisting of an emitter-base transmission line and 
a collector-current generator will be considered (Fig. 24). 

If r, g and c are respectively the resistance, conductance and capacit- 
ance per unit length of the line, and 1 is the length of the line, then by 

Fig. 24 — Transmission line and output-curr-ent generator representing idealised 


using the normal methods of network analysis, the general equations 
for the line may be written: 

= v e cosh PI — i e Z sinh PI 

V e 


i e cosh PI— — sinh PI 


Several facts may be established from these equations. For example, 
r e and c e may be related to r, g, c and 1, and hence, by means of the 
physical theory of the transistor, to its physical dimensions and 
constants. This method of attack leads to the equation 

kT j? 

qle ' 

on which the approximate expressions for r e given earlier are based, 
where k is Boltzmann's constant, Tj is the junction temperature in °K, 
and q is the electronic charge. 
Alternatively, r, g, etc. may be dispensed with by relating them to 

Te = 

Page 89 


measurable transistor parameters, such as the current amplification 
factor and cut-off frequency. It is the latter approach which will be 
followed here, in order to discuss the current amplification factor in 
grounded base. The treatment is limited to an idealised transistor, and 
in practice the parasitic elements, particularly Cdep and nab- , necessitate 
some minor modifications. 

Classical Approach to Current Amplification Factor 

The classical approach to the effect of frequency on the current ampli- 
fication factor will now be summarised. 

The fundamental equations for the transmission line give the current 
amplification factor as 

_ * 
cosh PI 

This expression gives rise to the diagram of Fig. 25, in which, for a 
series of frequencies, the imaginary part of a (lm(a) } has been plotted 
against the real part of a (Re(a) }, and the points joined up to form a 
locus. The point at which the magnitude of a (proportional to the 
length of the vector from the origin to the curve) has fallen by 3dB 
from its low-frequency value is marked on the locus. The frequency 
corresponding to this point is called the cut-off frequency f a , and this 
can be related clearly to r, g etc. and thence to the physical dimensions 
and properties of the transistor under consideration. The expression 
for a may then be written as 


~-coshV(2-44jf/f a ) 
and c e also may be expressed in terms of f a . 



The classical approach just outlined, which culminates in the definition 
of an arbitrary frequency characteristic f a , leads to several major 
difficulties : 

The expression for the a of an alloy-junction transistor (Eq. 1) is 
not particularly simple. For drift-field transistors, the transmission-line 
equations are more complicated and the corresponding expression for 
a is completely unmanageable. 

Drift-field transistors have different loci for a according to the 
strength of the drift field. The direct significance of f a , illustrated by 
Fig. 25, is therefore lost. 

At relatively high frequencies, f a is extremely difficult to measure, 
and almost impossible to measure with any accuracy. 

Page 90 


The relation between f a and other components of the equivalent 
circuits, particularly 

_ 1-22 
27if a r e 
is reasonably simple for alloy-junction transistors, but considerably 
more complicated for drift-field transistors. 

The relation between f a and the grounded-emitter cut-off frequency 

Frequency increasing 

Fig. 25 — Locus of a for increasing frequency 

f' a is not straightforward, particularly for drift-field transistors. 

These difficulties have become more apparent over the past few years, 
and a number of proposals have been put forward for overcoming them. 

High-frequency Parameter fi 

The present situation will now be outlined by discussing the difficulties 
in turn. 

A relatively simple approximate expression has been found for a, 
which holds for all practical values of drift field. This is 


a " l+jf/f« 
where <j> is a constant, the value of which depends on the drift field. 

A frequency having more direct significance than f a is that at which 
the real part of a is 0-5. 

A frequency which is much easier to measure than f a is that at 

Page 91 


which the magnitude of a' has fallen to one. This is called ft. 

The two frequencies at which Re(a) .= 0-5 and |a'| = 1-0 are for 
practical purposes identical. That is, when ft is measured, at which the 
magnitude of a' is one, what is in fact obtained is the frequency at which 
the real part of a is one-half. From Eq. 1, it can then be shown that 

For alloy-junction transistors (which have zero drift field) $ = 0-22, 
and thus f a = 1 -22ft . For alloy-diffused or other drift-field transistors, 
0-22 < <j> < 1-0. 

The relation between Cb'e and the frequency ft is now simple and, 
for all transistors, is independent of the value of the drift field : 


27Tftr e 
A simple relation also holds between ft and f' a , namely 

f _ ft 

From Eq. 1, it can be shown that |a'| falls at a rate of 6dB/octave 
(20dB/decade) at frequencies several times greater than f a . 

Relative Importance of ft and f a 

The alpha cut-off frequency f a has served well in the past to characterise 
the high-frequency behaviour of transistors, but it was chosen rather 
arbitrarily and suffers from some drawbacks. The introduction of ft, 
the frequency at which | a' J = 1, has eased the situation considerably. 

For certain special purposes, such as the discussion of the detailed 
behaviour of a, or the consideration of high-frequency noise, it is 
advantageous to retain f a , supplemented by ft . For many purposes, 
however, ft can be used by itself in place of f a . In particular, ft will 
come into greater prominence in connection with alloy-diffused 
and other drift-field transistors. 

Page 92 



Limits are set to the operating conditions in the interests of the user. 
The limits provide the circuit designer, whose experience is limited to 
comparatively small quantities of transistors and short operating 
periods, with ratings based on systematic life-testing of samples from 
the whole of the production. 

If the limits are exceeded the transistor may fail immediately; 
alternatively there may be a gradual loss of performance, and the life, 
normally very long, may be shortened appreciably. 


Transistor ratings are absolute. The ratings must never be exceeded. 

The circuit designer has to ensure that no transistor is operated 
outside the limits under any condition of operation and for any period 
of time. In arriving at the operating conditions, he must take into 
account: transistor spreads; variations in supply voltage; component 
tolerances; ambient temperature; and any other relevant conditions. 

It is not safe to assume that one rating may be exceeded provided 
a reduction is made in some other rating. If the collector voltage is low, 
the collector current cannot necessarily be increased up to the maximum 
permitted by the rated collector dissipation; in so doing, the collector- 
current rating itself might be exceeded. 


In the past, rather large safety factors were allowed in drawing up 
transistor data, because the users' and manufacturers' experience of 
transistors could not compare with that built up with valves over a 
great many years. Safe limits were adopted, and these were increased 
as more experience became available. 

This phase is now largely over. As a result of extensive life-testing, 
the ratings of the first transistors to be introduced have been increased. 
Newer transistors are rated as realistically as possible from the start. 

The existence of a limiting value may mean no more than that, to 

Page 93 


date, the region beyond the limit has not been investigated. But as 
time has gone on, more and more of the ratings have come to represent 
real limits of operation, beyond which performance and life are known 
definitely to be impaired. 


Most of the current flowing during the normal operation of the tran- 
sistor consists of an emitter-collector current. 

During its passage through the transistor, the emitter-collector 
current flows through regions of differing resistance, in which it exerts 
a normal heating effect. Little heat is generated at the emitter junction, 
which is of low resistance, and most of the heat appears in the region of 
the collector junction, where the voltage gradient is high. 

Provided the base power is small, the total dissipation in the transis- 
tor can be taken as being virtually equal to the collector dissipation 
V ce x I c . When the base power is not small, 

Ptot = Vcelc+Vftelb. 

The heat developed at the collector junction increases the junction 
temperature, the rise in temperature generally being greater for a 
smaller than for a larger junction. 

If the junction temperature is allowed to rise indefinitely, the transis- 
tor will fail catastrophically. 

At junction temperatures not sufficiently high to overheat the 
transistor, there may be long-term effects on life and reliability. 

The junction temperature is therefore rated at a certain maximum 
value, Tjmax. 

Effect of Ambient Temperature 

The heat generated inside the transistor has to be conducted away to 
the casing and absorbed by the surroundings. In continuous operation 
the transistor must reach a thermal equilibrium with its immediate 
surroundings, in which the collector junction remains at some steady 
temperature above the ambient value. 

The junction temperature of the transistor therefore depends on 
the ambient temperature T am b as well as on the collector dissipation. 
If the ambient temperature is high, the dissipation must be reduced, 
in order that the maximum junction temperature shall not be exceeded. 

The total permissible dissipation is calculated from 

Tjmax— Tamb 
Ptot ^ Pc = q » 

Page 94 



Ptot = total dissipation (~ collector dissipation p c ) 
Tjmax = maximum junction temperature 
Tamb = ambient temperature 

= thermal resistance, expressed as the rise in junction tem- 
perature above T am b for every unit of power dissipation. 

In this equation, the maximum junction temperature Tjinax is a limiting 
value and 6 is a constant. 

The permissible dissipation is therefore found from the maximum 
ambient temperature at which the equipment will be required to 
operate. For the OC71, Tjmax is 75°C for continuous operation, 


Fig. 1 — Permissible collector dissipation of OC71 plotted against ambient 


and 6 is 0-4°C/mW. If the equipment is required to operate up to 
temperatures where the air surrounding the transistor reaches 45°C, 

p cmax = __ 

= 75mW. 

This information is most conveniently given by a straight-line graph, 
such as Fig. 1, which is for the OC71. 

For a power transistor, more efficient means are provided for re- 
moving heat. For example, the effective or total 8 (from junction to 
ambient) of an OC23 when mounted on a suitable heat sink may be 
only 7-5°C/W. For a maximum junction temperature of 90°C, 


PtOt — tTc 

= 6W 
at a maximum ambient temperature of 45°C. 

Efficient Removal of Heat: the Heat Sink 

Heat is removed from the transistor by a combination of conduction, 
convection and radiation. The transistor manufacturer arranges for 

Page 95 


the collector junction to be in the best possible thermal contact with 
the casing. 

A further improvement can be obtained by making thermal contact 
between the casing and a metal plate, from whose surface heat is 
removed by convection and radiation. This plate may take the form 
of a heat sink or a clip-on cooling fin. Often the chassis forms the heat 
sink, though the effectiveness of this arrangement depends on the 
chassis temperature. 

Unfortunately, substances which are good conductors of heat are 
also good electrical conductors. In high-power transistors the require- 
ments of good heat conduction and electrical insulation are reconciled 

T *3-0°C/W 

Junction temperature 


» 0-5°C/W* 

Mounting base temperature 


e i 

= 0-2°C/W + 


Chassis (heat sink) temperature 



r Ambient temperature 

mounted on chassis without 

Fig. 2 — Thermal resistances of OC22, OC23 and OC24 power transistors 

as follows. The collector junction is connected electrically as well as 
thermally to a metal casing. Where no potentials exist on the heat 
sink which might affect the collector voltage, the transistor is bolted 
directly to the heat sink by means of two holes in the mounting base. 
Otherwise the transistor is insulated electrically from the heat sink 
(e.g., the chassis) by a mica washer and two bushes. Because the 
thermal insulation is as thin and as large in area as possible, this 
method of mounting affords the minimum resistance to the flow of heat. 

For a power transistor, 6 must be regarded as the total thermal resist- 
ance which is composed of three terms (Fig. 2). 6 m is the thermal 
resistance between the junction and the mounting base. 6j exists 
between the mounting base and the chassis (heat sink). 6h is the thermal ' 
resistance between the chassis (heat sink) and the surroundings. So 

__ Tj — T a mb 

Ptot ~~ Gm-f ei+Oh' 
This equation expresses mathematically the information given by the 
derating characteristic (Fig. 3). 

Page 96 


6 m is a function of the construction of the transistor; the manufac- 
turer keeps this as low as possible, and the circuit designer or transistor 
user can do nothing further to reduce it. The value of 6 m is 3-0°C per 
watt for the OC23. 6 t depends on the electrical insulation provided 



° 15- 

OC22, OC23 and OC24 





60"C/W ^N. 



Fig. 3— Permissible total dissipation of OC22, OC23 and OC24 plotted 
against ambient temperature for various 6h- 

between the transistor and the chassis. With the transistor insulated 
from the chassis, 6i is not more than 0-5°C per watt for the OC23. 

6 h depends on the heat sink provided by the user. It is determined 
mainly by the dimensions, position, material and finish of the heat sink. 
For example, an air-cooled heat sink may give a 6h of 4°C/W. 

0h can be determined for any heat sink from the temperature Tmb 
of the mounting base as measured by a thermocouple. Then 

6 1 mb -I amb Q 
h = Oj. 


The following example illustrates the temperatures at various 
points of an OC23 power transistor with mica insulation for p c = 4W, 
h = 4°C/W, 6 m = 3-0°C/W and 8i - 0-5°C/W. 
Junction temperature = 90 °C 
Mounting-base temperature = 90— (4x3-0) = 78 °C 
Chassis (heat sink) temperature = 78— (4x0-5) =76°C 
Permissible ambient temperature = 76— (4x4) = 60 °C. 
The suitability of any design can be checked by measuring the 
mounting-base temperature with a thermocouple, with the equipment 
operating at the required total dissipation and maximum ambient 
temperature. The result is compared with a graph of maximum 
total dissipation versus mounting-base temperature (e.g. Fig. 5, p. 99). 
If the point lies above the line, the design is unsatisfactory, and the 
dissipation must be reduced or the heat sink improved. The total 
dissipation is interpreted as the maximum reached by any transistor 
of the type in question. 

The effectiveness of the heat sink is partly determined by its position 

Page 97 


relative to other objects. Blackening assists cooling by radiation if 
there are no other hot surfaces in the vicinity. The effect of obstacles 
to cooling by convection should be taken into account. 





—J U-5-0 



3-9010-05 -*J U- 

3-10+0-05 — h U~ 

All dimensions in mm 


395 max - 



Fig. 4 — Dimensions of mounting base, lead and mica washers, and insulating 
bushes for OC28, OC29, OC35 and OC36. 

Good thermal contact is required between transistor and heat sink . 
The transistor should be bolted down evenly and the heat sink should 
be flat. 

For minimum thermal resistance between the case and the heat 

Page 98 


sink, the washer should be smeared with silicone grease (insulating 
quality). The edges of the mounting holes in the chassis should be 
free from burrs or thickening, which could puncture the washer or 
cause uneven contact. 

Nothing should be done to impair existing arrangements for re- 
moving heat from the. transistors or from neighbouring parts of the 
equipment, when servicing transistorised equipment. 

An example of a dimensioned drawing of the washers and insulating 
bushes is given in Fig. 4. 


In the past, the heating effect of the collector current has been a useful 
starting point when drawing up the collector-current ratings. 

If the usual working voltage of the transistor as an amplifier is 
x volts, and the maximum collector dissipation y watts, most require- 
ments are met by a rating of y/x amps for the continuous collector 


IC23 anc 



Fig. 5 — Permissible total dissipation of OC22, OC23 and OC24 plotted 
against mounting-base temperature. 

current. This reasoning may be confirmed by systematic life-testing 
of transistors at the current in question. Such a rating is satisfactory for 
small-signal circuits with the transistor biased into the linear portion 
of its transfer characteristic. 

However, when the transistor is operated in a bottomed condition, 
as in a flip-flop or other switching circuit, the collector current that can 
be passed before the dissipation rating is exceeded may be very large. 
Such a high current can be damaging in itself, apart from its heating 
effect. To meet the requirements of such circuits, the rating is extended 
upwards by means of life-testing. Finally a real limit is found, or a 
rating established which covers all the proposed applications. 

In practice the designer may have to restrict the collector current, 
whether peak or continuous, to less than the rated value, in order 
to remain within the permissible dissipation or limit distortion. 

Page 99 



A large amount of laboratory and field experience goes into the 
drawing up of the voltage ratings. The ultimate limit to the voltage 
ratings, however, is set by the avalanche effect. Avalanche multiplica- 
tion of the current carriers takes place in the collector depletion layer 
if the voltage across the layer exceeds some critical value. The critical 
voltage depends on some operating conditions and not on others. It is 
determined by the physics of the transistor, and depends on such 
factors as current density; but it has the same value whether the 
applied voltage is peak or direct. 

Increasing forward 
base current 

Fig. 6 — Usual form of output characteristic 

Fig. 7 — Output characteristic plotted to higher voltages 

Avalanche Characteristics 

The output characteristic of a transistor in grounded-emitter connec- 
tion is usually given in the form shown in Fig. 6. These curves are 
plotted for increasing forward base current. 

If the characteristics for forward base current are plotted for higher 
collector voltages, the curves shown in the left-hand part of Fig. 7 are 
obtained. As the collector voltage is increased, the slope resistance of 
the characteristic falls, until a region is reached where only a slight 
increase in collector voltage is observed up to very high values of 

Page 100 


collector current. This is called the avalanche region, the voltage across 
the transistor being the avalanche voltage. 

Fig. 7 also shows the characteristics for increasing reverse base 
current. As the collector current is increased, the characteristic turns 
back on itself. The voltage across the transistor falls rapidly until the 
avalanche region is reached. 

These curves are for current drive to the base (each of the curves 
being plotted for some constant value of base current), but the 
characteristics for voltage drive are similar. 

The avalanche characteristics are of particular relevance to switching 
circuits. In these circuits the transistor is normally switched between: 
the on or bottomed condition, in which the forward base current is 
high ; and the off condition, in which a pulse is applied in a direction 
which tends to cut the transistor off, and reverse base current flows. 


sistor c 


-20 -30 -40 



Fig. 8 — Collector current plotted against absolute maximum collector- 
emitter voltage. 

A voltage rating curve for a switching transistor is given in Fig. 8. 
This is for the transistor cut-off. If the collector-emitter voltage is 
allowed to exceed the values shown, catastrophic failure may occur. 
The precautions required to prevent failure depend on whether the 
load is resistive or inductive. 

Resistive Load 

There is no great difficulty in keeping the working point within the 
permissible area of operation with a purely resistive load. The resistive 
load line should be drawn, in such a way that it does not intersect the 
avalanche characteristic, as shown in Fig. 9. 

Inductive Load 

The behaviour of the circuit is more complicated when the load is 
inductive, as in relay-switching circuits, d.c. converters and audio 

Page 101 


output stages. On switching the transistor off, the rapid decrease in 
collector current induces a voltage 

T di 

v = L— - 


across the inductance L. Since the current will still be very high during 
this period, a condition of high current, high voltage exists. If the 
voltage is not limited by some circuit device (e.g. a catching diode), 
the working point may intersect the static characteristic for the value 
of reverse base current used (Fig. 7). If this condition is allowed to 
occur, the rate of decrease of collector current is given by 

di _ Vx 

dt ~ L 
rather than by the normal transistor switching time. The time taken 
to switch off may then be comparatively long (perhaps several milli- 
seconds, depending on the L/R time-constant); during this period the 
transistor is operating in a high-dissipation region and may be damaged 
or even destroyed. 

Fig. 9 — Correct and incorrect resistive load lines and breakdown characteristic 

This condition may be avoided by choosing a suitable transistor for 
the application and by correct circuit design. 

Softening Voltage 

The fall in slope resistance before the avalanche region is reached is 
referred to as softening. The softening voltage is defined as the voltage 
at which the slope resistance falls to some arbitrary value, and is a 
function of the external base-emitter impedance. While it is permissible 
to operate in the soft region of the characteristics — provided precau- 
tions are taken to see that avalanche cannot occur — the attendant 
non-linearity usually leads to excessive distortion. This effect is taken 
care of in circuit design by means of a characteristic which shows the 
permissible peak collector voltage plotted against the base-emitter 
impedance. The permissible voltage falls as the impedance is increased. 

Page 102 



After the characteristics of the semiconductor diode have been out- 
lined, the various types, such as point-contact, junction etc. will be 


Semiconductor diodes offer combinations of characteristics not 
available with thermionic diodes, and they have a number of other 

Life will be long, if the ratings are observed, as there is no heater to 
fail, no cathode coating to lose emission, and no vacuum to soften. 

Hum cannot occur. The saving of heater power may be significant in 
series-operated equipment. The absence of a heater favours the semi- 
conductor diode in equipment where only diodes are required, or 
where a heater supply would require long leads. 

Small size and weight, and the fact that no holder is required to 
connect it in circuit, allow the semiconductor diode to be combined 
with other components. Detector diodes which do not dissipate 
appreciable amounts of heat can be incorporated in the coil unit. 

Interelectrode capacitance is low, and this is an advantage if the 
load also has a low capacitance. 

These advantages bring with them some drawbacks which, in certain 
applications, make the semiconductor diode unsuitable. 

Whereas ideally a thermionic diode passes no current in the reverse 
direction, appreciable reverse current flows through the semiconductor 
diode, particularly at higher voltages and temperatures. 

The characteristic of the semiconductor diode is noticeably tempera- 
ture dependent in both the forward and reverse directions. 

Semiconductor diodes are hermetically sealed to protect them from 
atmospheric moisture and soldering contamination. The envelope 
may be metal, glass, or glass encased in metal. 


The semiconductor diode is represented by the symbol shown on the 

Page 103 


right of Fig. 1. The 'bar' is connected with the polarity of a cathode 
and the 'arrow head' with that of an anode. 

The thermionic diode on the left of Fig. 1 shows this polarity. 

The cathode is the positive output electrode in a rectifying circuit. 

Fig. 1 — Polarity of semiconductor diode 

The smaller semiconductor diodes are made in either a double- 
ended or a single-ended construction (Fig. 2). In the double-ended 
construction, a coloured band is placed on the end of the body nearer 
to the cathode; alternatively the whole of the tip may be coloured. In 
the single-ended construction, a coloured spot is placed on the same 
side of the body as the cathode. 


The form of the characteristic of the semiconductor diode can be seen 
from Fig. 3. The dotted curve shows what is predicted from theory. 
The characteristics of the various types of semiconductor diode will 


Y— - k - — Y 

Fig. 2 — Electrode connections of small double-ended and single-ended diodes 

diverge to a greater or lesser extent from the theoretical, but the full- 
line curve can be taken as showing their general shape. 

Forward Characteristic 

The forward characteristic at first glance resembles that of a thermionic 
diode, but there are several important differences. The characteristic 
for the semiconductor diode rises comparatively fast, corresponding 

Page 104 


to a lower forward resistance. The forward current for a voltage drop 
of JV to IV ranges from a few milliamps for certain types to tens or 
hundreds of milliamps for others. 

At low currents the characteristic follows an exponential rather than 
a three-halves power law. At high positive voltages, well beyond the 
working range, the characteristic approaches a straight line; there is 
no saturation effect. 

Reverse Characteristic 

A negative voltage produces a small negative current. The reverse 
characteristic is plotted in the opposite quadrant to the forward 








Fig. 3 — General form of forward and reverse characteristics. Dotted line 
theoretical, full line practical. 

characteristic, the scale of the reverse current being in microamps, 
while the forward current is usually plotted in milHamps. The reverse 
current is sometimes referred to as the leakage current, and is similar 
to the leakage current through the collector junction of a transistor. 

The inverse current increases rapidly at first until the low-voltage 
'knee' is reached. Then the current increases gradually until, at a point 
called the turnover voltage, there is a rapid increase and the character- 
istic may turn back on itself. If the current is not limited in the turnover 
region, and the maximum dissipation is exceeded, the diode overheats 
and is permanently damaged. 

The rating for the peak inverse voltage is of fundamental significance 
and should be observed strictly. Excess voltages lasting even a few 
microseconds may be disastrous. 

Zero Region 

The characteristic passes through a true zero (no voltage, no current). 
In measuring circuits, compensation for the starting current, which is 
often necessary with a thermionic diode, is not required. 

In the zero region (±10mV) the characteristic is approximately 
linear and, as the forward and reverse resistances are nearly equal, 

Page 105 



the detection efficiency at this level is low. The detection (or rectifica- 
tion) efficiency in a specified circuit at a given peak input voltage is 
_ d.c. output voltage xlQ0 ^ 
peak input voltage 


All measurements and ratings only apply to the temperatures at which 
they were made or specified. At higher temperatures, it is necessary to 
derate the permissible forward current and sometimes the maximum 
inverse voltage. 

A further derating to the forward current is required if appreciable 
inverse voltages occur during any part of the cycle. Although the 

I 20- 



Fig. 4 — Mean forward current (averaged over any 50 millisec period) 

plotted against peak inverse voltage for the OA70 at ambient 

temperatures up to 25°C. 

reverse current is much lower than the forward current, the high 
inverse voltage may give appreciable dissipation in the reverse direction. 

Derating information for the OA70 is given in Fig. 4. At ambient 
temperatures above 25°C (up to the permitted maximum of 75°C), 
the maximum mean forward current should be multiplied by 25/T am b • 
Similar information is given for other types in the published data. 

Disregarding the rating and derating information may cause the 
diode to run away and break down. 

It is not to be assumed that the semiconductor diode is excessively 
sensitive to temperature and inverse voltage. The dangers have been 
mentioned to draw attention to the deratings required. This aspect of 
circuit design should be given as much attention as is given to correct 
h.t. and l.t. voltages in ordinary valve practice. 

Page 106 



One extremely important aspect of diode performance does not appear 
on the d.c. characteristic curve. 

The semiconductor diode usually consists of a strongly p-type anode 
and a slightly n-type cathode. When the diode is conducting in the 
forward direction, holes are injected into the cathode. On reversing 
the polarity, the holes remaining in the 'cathode' flow back into the 
♦anode'. Thus the current through the diode does not fall to the leakage 

-Forward current 

— Peak reverse currents 

Fig. 5 — Stored charge and reverse current 

value as soon as the voltage across it is reversed, but a pulse of current 
flows for a short time in* the reverse direction, until the current settles 
down to the leakage value (Fig. 5). 

This effect results from the storage of current carriers. The effect is 
normally referred to as hole storage. 

The peak inverse current is equal to the applied inverse voltage 
divided by the resistance of the external circuit. 

The diode behaves as if an additional capacitance were connected 
across the junction of its two elements. This is called the storage 
capacitance, to distinguish it from the interelectrode or reverse capaci- 
tance of the diode biased in its reverse direction. The concept of storage 
capacitance is, however, not very helpful, as the capacitance is non- 

Carrier storage may be expressed quantitatively in terms of the 
inverse current flowing in the circuit a certain time after the forward 
current has been removed. For a typical OA86, when a forward current 
of 30mA is removed and an inverse voltage of —35V applied to the 
diode through a resistance of 2-5kQ, the inverse current decays to 
380(jiA after 0-5 microseconds and to 36(jlA after 3-5 microseconds. 
The OA86 is suitable where low carrier storage is essential, as in some 
computer applications. 

Alternatively the carrier-storage characteristic may be quoted 
as the time required for the transient current to fall to a certain value. 
This time is the 'recovery time'. For the OA10, when a forward current 
of 10mA is removed, and an inverse voltage of 7V applied to the diode 
through 300£J, the time taken for the inverse current to decay to 0-5mA 
is typically 0*18[xsec. 

Page 107 


Two recovery times may be observed in practice. Most of the stored 
charge decays within a short time, but a longer term effect can be 
detected at low levels which is sometimes important. 

The recovery time is meaningful only if the circuit, the forward 
current, and the switching time of the pulse applied to the diode are 
quoted. It is misleading to compare recovery times without at the same 
time considering the conditions of measurement. 

There is a good deal of discussion at present as to the best method 
of measuring carrier storage. It is not really possible to infer the per- 
formance in every circuit from one measurement of carrier storage in 

Vf» Forward voltage 
V s =Transient forward voltage 

Fig. 6 — Turn-on transient forward current 

one standard circuit, but the information in the published data repre- 
sents the practical requirements as closely as possible. 

The recovered charge may be quoted in future. This is equal to the 
area under the reverse peak, the shape of the peak being determined 
by the circuit resistance. 

Turn-on Transient Forward Voltage 

During the build-up of stored charge in the base region, the resistance 
of the base decreases. 

When a pulse of forward current is passed through the diode, the 
forward voltage is initially higher by an amount V s than the steady 
forward voltage V f (Fig. 6). This effect is known as the forward 
recovery of the diode. For the OA10, the transient forward voltage V s 
is less than 0-2V for a forward current of 400mA and a rise time of 


The point-contact diode was the first of the semiconductor diodes to 
be developed. It is still of great importance in circuit design, and is 
firmly entrenched in a number of applications, where it offers a per- 
formance not available with newer forms of construction. It is unlikely 
to suffer the fate of the point-contact transistor, which has been 
rendered obsolete by the junction technique. 

Page 108 


The point-contact diode is the modern form of the crystal detector 
(cat's whisker) used in the simplest radio receivers, and was developed 
originally for operation at radar frequencies, where the low inter- 
electrode capacitance and short 'transit time' provided a great improve- 
ment over the thermionic diode. 

Structurally the point-contact diode consists of a springy tungsten 
wire which presses against a crystal of germanium. During manu- 
facture a current is passed between the whisker and the crystal to 
'form' the diode. A small region of p-type material is formed beneath 
the point of the whisker. Rectification takes place between this region 
and the remaining material which is n-type. 

Although the point-contact diode is effectively a junction diode of 
small interface area, there are considerable differences. At low reverse 
voltages the current in the point-contact diode approaches the 
theoretical value, but the slope of the characteristic thereafter rises 
up to the turnover voltage. Generally speaking the forward voltage 
across the point-contact diode is comparatively high. On the other hand, 
the reverse capacitance and usually the carrier storage are lower than 
for the junction diode. The temperature sensitivity of the two types is 

The point-contact diode is generally the most suitable of the semi- 
conductor diodes for r.f. detectors and mixers. 

High and Low Reverse Voltages 

Point-contact diodes can be produced with a variety of characteristics. 
Heavily doping the germanium reduces the maximum reverse voltage. 
Lightly doping the germanium makes for a higher reverse-voltage 

A heavily doped diode can operate efficiently as a detector up to very 
high frequencies but has a low reverse-voltage rating. At the other 
extreme, a p.i.v. rating of over 100V can be provided in this technique. 

The OA70 is suitable for operation as a detector up to lOOMc/s; 
the OA81 is a high voltage diode for general applications requiring 
the higher reverse-voltage rating and is suitable, for example, for 
shunting relay contacts (either to suppress sparking or to protect a 
transistor from voltage surges). 

The OA85 has a lower voltage drop than the OA81. The OA86, for 
which the maximum p.i.v. is 90V, has better hole-storage character- 
istics than the OA85 and is more suitable for computer circuits. 

What are essentially miniature versions of the OA81 and OA85 
are provided by the OA91 and OA95. 

A reverse-voltage rating intermediate between the two extremes 

Page 109 


(45V) is provided by the OA79. This diode is suitable for sound detec- 
tion, and is available in matched pairs (under the type number 2-OA79) 
for use in the discriminator or ratio detector in f.m. reception (Fig. 7). 

Applications of Point-contact Diodes 


Fig. 8 shows a typical video-detector circuit for operation at 30Mc/s. 

The OA70 should be used in this circuit; the recharging time for the 


Fig. 7 — Ratio detector using 2-OA79 

10pF capacitor is short when operating at 30Mc/s, and this diode has 
a low forward resistance (about 10mA will flow for 1 volt drop) which 
will allow a substantial charging current to flow. The lOOkti reverse 
resistance of the OA70 is more than sufficient to prevent the capacitor 


Fig. 8 — OA70 as video detector at 30Mc/s. Peak input voltage 5V, damping 

resistance 3kO. 

from discharging back through the diode, instead of through the 3-9kQ 

The rectification efficiency in this circuit is at least 54 % for an OA70 
at 30Mc/s, with a peak signal of 5V. 


An extended reverse characteristic is essential in some applications. 

For example, in a sound-detector circuit the 3-9kQ load resistor of 

Page 110 


Fig. 8 would be replaced by, say, 47k£2, and the choice of diode lies 
between the OA70, OA79 and OA81, depending on the peak inverse 
voltage which will be encountered and on the value of the load resistor. 

Detection in a transistor receiver is preferably performed by an 






f OA70 

OCAb /7> 



:£?• *4* 

56 kO 
'WW o 


Fig. 9 — OA70 as sound detector in transistor receiver 

OA70 and not by a transistor (Fig. 9). The final i.f. amplifier is an 
OC45 and the first a.f. amplifier is an OC71. 


The noise limiter shown in Fig. 10 requires a diode with a high reverse 

resistance, such as the OA81. A small current flows through the chain of 

Fig. 10 — OA81 as television noise limiter 

1MQ resistors and holds the diode in its conducting region. The diode 
therefore provides a path for normal audio-frequency signals. Inter- 
ference, however, drives the diode into its reverse-current region, where 
the high reverse resistance virtually open-circuits the signal path. 


The gold-bonded diode is structurally a point-contact diode in which 
the usual tungsten-wire contact is replaced by one made of gold. The 
result is effectively a small-area junction diode. The characteristics are 
a cross between those of the point-contact and junction diodes, the 
reverse capacitance being low and the forward current high. The great 

Page 111 


improvement in front-back impedance ratio, and the low voltage at 
which forward conduction starts, make the device of particular interest. 

The OA5 is in a single-ended construction. The forward-voltage drop 
is 0-4V at 10mA, while the capacitance of lpF is very little more than 
that of a conventional point-contact diode. The maximum inverse 
voltage is 100V at 25°C and 50V at 75°C. Some of the possible applica- 
tions are: in ring hiodulators for telephony; as a catching diode; and 
in transistor computing circuits. 

The single-ended OA7 and the double-ended miniature OA47 have 
essentially the same electrical characteristics as one another. They are 
for use with transistors operating at peak currents of up to 50mA in 
the 'logic' circuits of computers operating at pulse-repetition rates of 
up to lMc/s. 


A junction transistor consists of two diodes formed on a base wafer, 
whilst a junction diode consists essentially of one such diode. 

Comparison of Silicon and Germanium Junction Diodes 

The properties of the junction diode are determined largely by whether 
it is made from germanium or silicon. 

Silicon gives a considerably higher reverse-voltage rating, and is 
preferable for power rectifiers for medium and high voltages. German- 
ium gives a lower forward-voltage drop, and higher efficiency in a 
power rectifier for low voltages. 

A higher junction-temperature rating can be realised with silicon 
than with germanium, so that this, too, makes silicon more suitable 
for high-power rectifiers. 

As a rule the silicon diode has the lower reverse current, and the 
germanium diode the better high-frequency performance. 

Characteristics of the Junction Diode 

The forward volt drop of a junction diode is low — the forward current 
being about 100mA at 0-5V for an OA10. The reverse current remains 
low up to quite high voltages, but increases sharply if the maximum 
rated voltage is exceeded. The characteristics of most junction diodes 
approach the theoretical more closely than do those of point-contact 

In addition to the storage capacitance, the capacitance of the diode 
when biased in the reverse direction may be important when the diode 

Page 112 


is used with tuned circuits operating over a range of signal levels. The 
interelectrode or reverse capacitance usually varies inversely as the 
square root of the voltage. 

Both forward and reverse currents increase substantially with 
temperature. The reverse-voltage rating usually decreases with 
increasing temperature. 

Junction diodes are more robust than those manufactured by the 
point-contact technique and are generally able to stand higher levels 
of shock and vibration. 

The OA10 is designed to have sufficiently low hole storage for 
square-loop-ferrite coupling circuits, and in consequence has a corn- 

Fig. 11 — Four OA10 in low-voltage bridge rectifier. Peak current 700mA 

paratively low reverse-voltage rating (30V). This diode is suitable for 
low-voltage rectification (Fig. 1 1). 

The OA200 and OA202 are miniature general-purpose silicon diodes. 

Silicon junction rectifiers are available with very high reverse voltage 
ratings for rectification at currents of several hundred milliamps. 
Large metal-case rectifiers designed for bolting down to a heat sink 
commonly have rather lower voltage ratings but can rectify currents 
of 15A and upwards. 


A silicon rectifying junction has a more sharply defined 'knee' in 
its reverse characteristic than a germanium junction. Beyond a certain 
point the current increases sharply with voltage, the dynamic impedance 
in this region often being only a few ohms. 

The silicon Zener diode exploits this part of the characteristic and 
provides a nearly constant voltage over a reverse current range of tens 
of milliamps. It is used as a voltage stabiliser or voltage limiter. 

Two factors determine the usable current range. The lower limit is 
set by the fact that at low currents the slope resistance increases and 
the voltage falls. The upper limit is set either by the dissipation or by 
the maximum-current rating. 

Zener diodes possess a normal forward characteristic. 

Page 113 


Zener diodes can be specified as a series of types covering a range 
of nominal Zener voltages, the individual types having a narrow or 
wide tolerance on the particular values. Thus Zener diodes are available 
under the type numbers OAZ200 to OAZ207 and OAZ208 to OAZ213. 
The OAZ200-207 cover progressively higher nominal voltages from 
4-7V (OAZ200) to 9- IV (OAZ207) with a tolerance on the nominal 
voltage of approximately ±5%. The OAZ208-213 give a coverage 
from 4-2V (OAZ208) to 12-2V (OAZ213) with a tolerance of approxi- 
mately ±15%. 

Page 114 



The phototransistor is a junction transistor, manufactured in such a 
way that the inherent photoelectric properties are fully exploited. It is 
essentially a photodiode, in which the light current is amplified by 
normal transistor action. 


The mobile current carriers on the two sides of an isolated p-n junction 
will diffuse across the junction, until a counter-e.m.f. is built up which 
limits the further exchange of carriers. When a reverse voltage is applied 
to the diode, the barrier potential is increased correspondingly until 
equilibrium is again reached. Under these conditions, some carriers 
receive sufficient random energy to cross the barrier, and constitute the 
normal leakage current. This is the dark current. 

If light is allowed to fall on the junction, hole-electron pairs are 
created on both sides of the junction. The barrier potential sweeps the 
holes one way and the electrons the other. The current now flowing is 
the light current, equal to the dark current plus the photoelectric 


The phototransistor operates like a normal transistor into which light 
is allowed to enter. The hole-electron pairs generated by the incident 
light should preferably be created as near to the collector junction as 
possible, in order to minimise recombination. Hole-electron pairs 
stand very little chance of contributing to the light current, if they are 
generated more than one diffusion length (usually about 0-5mm) from 
the collector junction. 

With the base open-circuit (Fig. 1), the leakage current I co , which 
is only a few micro-amps, is transformed by transistor action into 
I'co, which is equal to (l+a')I C o. The dark current with the base open 
circuit is substantially equal to I' co . 

When light falls on the phototransistor, the photoelectric current 
Iph contributes an internal base current which, if the base is open 
circuit, is amplified by the factor (1+a') into a much larger current at 
the collector. The light current in this circuit is approximately 

Page 115 


(1 +a')I P h+r C o = (1 +«') (Iph+Ico). Provided the photoelectric current 
is large in comparison with I C o, the light- to dark-current ratio is high. 

Normally, the phototransistor is operated with a resistance between 
base and emitter, to ensure thermal stability at higher temperatures. 
The dark current (I' co ) is then reduced considerably below its value 


r eo .<*a')i eo 


Fig. 1 — Relation of I'co to l C o 

with open-circuit base, resulting in a greatly improved light- to dark- 
current ratio. 


The OCP71, which operates on the principles just described, is a 





Fig. 2 — Base connections of OCP71 

general-purpose device of p-n-p alloy-junction construction. The base 
connections are as shown in Fig. 2. 

Sensitivity and Response 

Maximum spectral response occurs at a wavelength of 1-55/u,, and 50% 
response is obtained at about 0-8[x and l-6(x. (The micron hx] is 10~ 6 of 
a metre, and is equal to 10,000A.) The peak response is in the near 

Page 116 


infra-red, but continues through the visible spectrum to the near 

Light falling on the emitter side of the crystal creates hole-electron 
pairs in the parts of the base not in the shadow of the emitter (Fig. 3). 
The chief contribution to the total light current is made by hole- 
electron pairs which have only a short distance to diffuse before reaching 
one of the junctions. The emitter junction of the OCP71 is smaller than 
the collector junction. The hole-electron pairs generated in an annulus 
('doughnut'), the inner and outer boundaries of which are formed by 
the emitter and collector, are therefore most effective. 

If the light is incident on the collector side of the crystal, the collector 
shades the annulus bounded by the emitter and collector. Hence fewer 

Base materiol 

Fig. 3 — Diagrammatic cross-section of OCP71 

hole-electron pairs are created, and recombination is greater, and the 
total light current is very much smaller. However, the moistureproof 
grease in which the assembly is coated diffuses the light, and the device 
is not critically sensitive to its direction. 

Maximum current is obtained when the light is allowed to fall on the 
side of the bulb bearing the type number (Fig. 2), in a direction perpen- 
dicular to the plane of the leads. The response is about 50 % of the 
maximum, when the light is incident at 90° to this direction in the 
horizontal plane. If the light falls vertically downwards, the response is 
still 30% of the maximum. 

Output Current 

The output (Ic/Vc) characteristic of the OCP71 is plotted for a number 
of intensities of incident light. The intensity, which is specified in foot- 
candles ( = lumens/ft 2 ), has to be calculated in order to interpret this 

The amount of light (in lumens) falling on any surface is given by 
Candlepower of source X Area of surface 
(Distance of surface from source) 2 
For example* if a 40W tungsten-filament electric lamp (which gives 

Page 117 


approximately 30 candlepower) is placed 120mm from a surface 7mm 2 
in area, the flux at the surface is 

30X7 AAK! 

-^-^ 0-015 lumen. 

An OCP71 phototransistor has an effective sensitive area of about 
7mm 2 . If used with a 40W unfocused lamp, as in this example, the 
OCP71 will pass a current of 4-5mA, since the sensitivity is about 
300m A/lumen. 

The above example using unfocused light does not give a very clear 
idea of the high sensitivity of the OCP71. A more vivid illustration is 
provided by a 2\N pea-lamp, under-run from a supply of 1|V. If the 
barely glowing filament is several centimetres away from the OCP71, 
and the light is focused by a simple glass lens on to the sensitive area, 
the current is at least 5mA. 

Collector Dissipation 

The junction temperature of the OCP71 must not be allowed to exceed 
65°C. Junction temperature, ambient temperature and collector dissi- 
pation are related in the usual way. The thermal resistance is 0-4°C/mW, 
and the permissible collector dissipation for any maximum ambient 
temperature is found from 

, __,. Tjmax— Tambmax 
p c max(mW) = -* — . 

To take a specific example, if the circuit is intended to operate up to 
an ambient temperature of 45°C, 

65—45 _ A ... 

p c max = — -r— — = 50m W. 

Precautions should be taken where necessary to prevent thermal 
runaway, and d.c. stabilisation should be provided as required (Chap. 
7). Without some form of d.c. stabilisation or an external base resis- 
tance, the ambient temperature must be restricted to 25°C, at the 
maximum permissible collector voltage for the OCP71 of 25 V. 

Circuit Design 

The circuits which may be employed are extremely simple, and may 
amount to nothing more than the phototransistor connected in series 
with a relay coil and a d.c. supply of 12 to 18V, with the base left 
unconnected. If it is required to operate the phototransistor over a 
wider range of temperature or from a high voltage, as in most industrial 
applications, a resistor should be connected between base and emitter, 
to reduce the dark current and hence improve the thermal stability. 
A suitable value may be of the order of 5k£X 

Fig. 4 shows a basic circuit for d.c. (unmodulated light). This circuit 

Page 118 


is adequate for 'on-off 'applications. With a base-emitter resistor of 
5kQ, the light- to dark-current ratio may be, for example, 480 at 25°C 
and 20 at 45°C. Without this resistor, the corresponding values may 
be 90 and 9-5. The resistor should preferably be an n.t.c. type; the 
exact value depends on the maximum ambient temperature and the 
light level. 

Fig. 5 includes an OC201, which is used as a simple d.c. amplifier 
following the OCP71 to give extra sensitivity. This circuit has tempera- 
ture compensation. 

Maximum dissipation in the transistor occurs at half of full drive. 
Maximum power is available to operate the relay when the transistor 

OA8I ^ _ 

Fig. A — Basic circuit for unmodulated light 

Fig. 5 — Temperature-compensated circuit with d.c. amplifier 

is bottomed, under full drive. For example, consider a phototransistor 
operating from a 20V supply. A relay of 5kQ is in the collector. Sufficient 
light is available to bottom the phototransistor and 4mA flows in the 
collector. Thus : 

Power in the load ~ 4mA X 20V = 80m W; 
Max. power in the phototransistor on switching 

4 A 20 17 

~ -mAx-V 
2 2 

20m W; 

Phototransistor dissipation when 'bottomed' 

~ 0-15Vx4m'A = 0-6mW. 

To ensure reliable operation, it may be necessary to choose a relay 
which will pull in at a power rather lower than four times the maximum 
power in the transistor. 

Where temperature stability is important and a 'chopped' light source 
is available, the circuit of Fig. 6 can serve as a basis for design. The 
base is returned to the emitter through an inductance, which is prefer- 
ably parallel-tuned to the light-modulation frequency. For the modula- 
tion frequency, the base is essentially open-circuit, and maximum 

Page 119 


amplification of the light signal is obtained. For d.c. (dark current, or 
current due to unwanted background illumination), the base impedance 
is so low that the amplification of this current is reduced to a minimum. 
Stabilisation is provided by the potential divider and by the bypassed 
resistance in series with the emitter. 

Fig. 6 — Basic circuit for modulated light 

The cut-off frequency for a phototransistor is the modulation 
frequency at which the gain is reduced 3dB below the value with un- 
modulated light. The typical value is 3kc/s for the OCP71 in the circuit 
shown in Fig. 6. 

If the emitter resistor is unbypassed, a.c. feedback will be introduced. 
The overall gain will be considerably lower, but the frequency response 
will be considerably extended. 


Special features of the OCP71 are its small size, low operating voltage, 
high sensitivity, relatively large sensitive area, robust construction, 
ability to respond to infra-red radiation, and quick response. 

The OCP71 is suitable for a variety of industrial-control and other 
applications such as: 

Photoelectric counters 

Speed measurement 

Liquid-level controls 

Edge detection (in paper making, textiles, and belting) 

Burglar alarms 

Door opening 

Curve followers 

Smoke detection 

Industrial on/off controllers. 

Page 120 



In designing a transistor amplifier, it is more convenient to consider 
the early stages as current amplifiers rather than as voltage or power 
amplifiers. First the output stage is designed and the drive requirements 
are determined. The driver stage is then designed, and a sufficient 
number of amplifier (or preamplifier) stages added to give the required 
The stages of the amplifier fall roughly into three groups : 

(a) Low signal (1st stage and possibly 2nd) 

(b) Medium signal (middle stage or stages) 

(c) Large signal (output stage, and possibly the driver). 

Case (b) is the simplest, and (a) and (c) are extreme cases of (b). 

The power output and driver stages are covered chiefly by the two 
following chapters. This chapter will be devoted primarily to the 
amplifier stages, but will include some general remarks which apply 
to audio amplifiers as a whole. 

The emphasis will be on RC coupling, as this is almost always used 
in practice for audio amplifier stages, and is most convenient for those 
who want to design their own circuits. 


Common-emitter connection is almost always used for low-level a.f. 
amplifiers, because of the high power gain obtainable in this con- 
figuration. Either transformer or RC coupling is suitable. 

RC coupling is preferable for a multistage amplifier, since otherwise 
instability may be introduced by magnetic coupling through the 
transformers. Also, the design is simpler for RC coupling, the com- 
ponents are readily available, and the final equipment is lighter and 
less bulky. 

Even when an additional stage is required to achieve the same gain 
with RC as with transformer coupling, the cost of the equipment is 
not likely to be increased. 


By means of d.c. stabilisation, satisfactory performance can be assured 
for any transistor of a given type over a specified temperature range. 
Reference should be made to Chapter 7, where the stability of various 

Page 121 


circuits and the influence of each component are described. The present 
section gives a practical method for calculating the maximum and 
minimum collector currents over the range of operating temperature. 

As was shown in the earlier chapter, the preferred biasing circuit for 
common-emitter connection is that of Fig. 1. The supply is assumed 


v cc J 




, 1. -. 

,R e 


Fig. 1 — Conventional potential-divider and emitter-resistor circuit 
Fig. 2 — Two-battery circuit equivalent to preceding circuit 

to be a single battery, as is normal practice. The two-battery circuit, 
which is a more basic biasing arrangement, is shown in Fig. 2. Both 
these circuits are reducible to the same equivalent circuit. 

From Fig. 2, 

Vbb = IbRb+Vbe+IeRe ... (1) 

where Rb is the resistance in series with the base. For the one-battery 


Rb = 




VccR2 . 
R1 + R2 

The collector current is given by 

Ic = Ico+ale 
and since numerically 

Ic = Ie— lb , 

lb = Ie(l— a)— Ico- 

Substituting for lb in Eq. 1 and rearranging gives 

T _ Vbb — Vbe +IcoRb ^ Vbb — Vbe +IcoRb 
6 ~ R e +Rb(l-a) ~ Re+Rb/a' 

Vbb , Rb and R e are under the control of the circuit designer, but 
variations in battery voltage, and resistor tolerances, necessarily 


Page 122 


introduce a spread in emitter current. Vbe , a' and I C o are properties 
of the transistor, and of these I co and to a lesser extent Vbe are tempera- 
ture dependent. Hence transistor production spreads and changes in 
temperature introduce a further spread in emitter current. 

Maximum and minimum values of emitter current can be found by 
substituting the appropriate extreme values in Eq. 2. 

The changes in Ico and Vbe both increase the emitter current with 
temperature. Maximum emitter current occurs at maximum ambient 
temperature, and conversely. 

Vbe decreases with temperature at the rate of roughly 2mV/°C, so 
that the value given by the characteristic curves, which normally are 
plotted for an ambient temperature of 25°C, must be corrected for 
other temperatures. 

Ico increases exponentially with temperature, and its value can be 
found from a graph, such as Fig. 3 for the OC71. 

Operating Current and Voltage 

The maximum and minimum emitter currents must be such that 
sufficient peak voltage and current is available to drive the next stage, 
and the values of Rb and R e have to be chosen accordingly. The circuit 

" 16 ■ 

::::::::::_: ^ ;?_:: : 



25 30 35 40 45 50 55 

Fig. 3 — Ratio of l co at a given junction temperature to its value at 25°C. 

should not be designed for peak currents or voltages greater than the 
required values, or the gain will be unnecessarily low. 

The OC70, OC71 and OC75 should not be operated in the region 
below 0-3mA, where the non-linearity of the a'/Ic characteristic 
produces excessive distortion. The current drive available is 

ic( P k) = Ic— 0-3 [mA]. 

There is no advantage in providing a higher current than required. 

The voltage drive available is given by 

Vc(pk) = V C c — Vknee — leRe — IcRc- 

Vknee , the knee voltage, is the minimum collector-emitter voltage 

Page 123 


for satisfactory transistor action. In class A amplifier circuits, the 
transistor is 'bottomed' when the base current is such that the collector- 
emitter voltage is equal to the knee voltage. Any further increase in 
base current does not influence the collector current. 


The collector current increases with junction temperature. Whether 
or not the increase in collector current causes an increase in dissipation 
depends on the circuit conditions. 

If the collector current of the transistor is varied, while keeping the 
values of the collector and emitter resistors fixed, the collector dissipa- 
tion is a maximum when the voltage across the transistor, V ce , is equal 
to half the supply voltage. The collector dissipation is then 

["XT! Vcc 2 

LRj 4(R C +R e )' 
If the operating conditions are such that V ce is greater than |V CC , 
then an increase in I c causes an increase in the collector dissipation. 
If V ce is less than |V CC , then an increase in I c causes a decrease in 

Circuits with V ce < ?V CC 

Most RC-coupled and some transformer-coupled stages come into this 
class. Such circuits are thermally stable because the dissipation decreases 
with increase in temperature. This property can be turned to advantage 
to obtain higher dissipations at normal ambient temperatures. 

To evaluate the maximum junction temperature, the lowest value 
of collector current should be calculated for the maximum ambient 

Circuits with V ce > -jVcc 

Many transformer-coupled stages and output stages come into this 

class. For such circuits two calculations are necessary. 


The maximum emitter current ( ~ I c max) is determined at the maxi- 
mum ambient temperature, as described under D.C. Stabilisation. 
The power dissipated in the transistor may be taken as 

pc = I c V ce , 
and the maximum junction temperature is found by inserting the 
maximum values of ambient temperature and collector dissipation in 
the equation 

Page 124 

Tj = T am b + 6pc , 

where 8 is the thermal resistance from the junction to the surroundings. 
Tj must not exceed the maximum permissible value. 


The circuit is thermally unstable if 


and the dissipation then increases to the limit, V C c 2 /4(R e +Rc), 
imposed by the circuit. If the increased dissipation results in a tempera- 
ture greater than the maximum permissible junction temperature, the 
transistor may be permanently damaged and possibly destroyed. 

dpc/dTj may be calculated from the following equations : 

^-[v M - 2 WR. + i W ].g 


dl_e = (Re+Rb) (dT~J " Iff 

dTj R e +R b /«' 

In the above equation, 

gj~ ^ 0-Q8I C o(Tj) 

(Ico(tj) being the value of I c0 at the junction temperature Tj), and 

dV be 




The a.c. load Rl on the transistor is formed by the following in 

(a) the output resistance r ce of the transistor 

(b) the load resistance R c in the collector circuit 

(c) the bias resistance Rb 

(d) the input resistance Rm of the following stage. 

J_ = J_ 1 J_ J_ 

Rl r C e Rc Rb Rin 

This equation can be evaluated more conveniently by substituting 
conductances for the reciprocals, thus 

Gl = gce+Gc+Gb+Gi n . 
Page 125 


The output current from the transistor, which is equal to a'ib , will not 
all flow into Gm , some of it being shunted off by the other conductances 
The current gain of the stage is therefore 

A, = «' 9* 

gee + G c + Gb + Gin 

The gain will be increased as G c and Gb are reduced, that is, as R c and 
Rb are increased. 

The effect of the coupling capacitance is discussed below. 


The coupling capacitance is considered to be connected between the 
input resistance Rm of the following stage and the source resistance 
R s of the preceding stage. R s will be formed by the collector load 
resistor R c in parallel with the high output resistance of the transistor, 
and is approximately equal to R c . 

A 3dB reduction in gain occurs at the frequency at which the reactance 
of the coupling capacitance is equal to Rm+Rc+Rb , that is, when 

The input resistance is typically lkQ, the collector resistor about 5kQ, 
and Rb may be only 2kQ. For a 3dB reduction in response at lOOc/s, 
and using these values, the coupling capacitance would be 

C= I F 

2tt. 100.8000 

~ 0-2(xF. 

Normally the coupling capacitance will be an order of magnitude 
larger than this to ensure good bass response, and values of 4, 8 and 
10[xF are commonly used. 

The polarity of the electrolytic coupling capacitor must be observed. 
The negative side of an interstage coupling capacitor is connected to 
the collector of the preceding transistor, and the positive side to the 
base of the following transistor (Fig. 4, p. 130), assuming that the 
transistors are p-n-p types. 

In an input stage, where the capacitor forms part of a base-emitter 
network, the negative side of the capacitor is connected to the base. 


The resistance R to be decoupled at the emitter is not that of the 
emitter resistor alone, but is given by 

Page 126 

R Rs Re 

where R s is the total source resistance feeding the base. 

As an example, a' can be taken as 50 and R e as lkQ. The output 
resistance of the transistor is in the region of 20 to 50kQ, hence the 
source resistance is substantially equal to the collector load resistance 
of, say, 5kQ. R e is large in comparison with R s /a', and the resistance 

a' 50 

Again a large capacity is normally used to ensure good bass response. 
The decoupling capacitance is usually 100}xF, the reactance of which 
is about 16Q at lOOc/s, 32Q at 50c/s, etc. Decoupling capacitors of 
250(xF are not uncommon. For a p-n-p transistor, the negative side of 
the decoupling capacitor will be connected to the emitter and the 
positive side to the positive supply line (Fig. 4). 


The h.f. response of the circuit is determined by the transistors, the 
cut-off frequency in grounded emitter, f' a or fp , giving a fairly good 
guide to the frequency at which a 3dB loss of gain occurs in practical 
grounded-emitter circuits. 

The l.f. response is determined by the coupling capacitances and the 
emitter decoupling capacitances. 

If the interstage coupling capacitors and emitter decoupling capacitor 
are added to the stabilised circuit of Fig. 1, the result is the basic 
RC-coupled amplifier stage of Fig. 4. 

In practice an amplifier consisting of a number of RC-coupled 
stages is designed initially to have as wide a frequency response as 
possible, using coupling capacitances much higher than those indicated 
by simple theory. Then the response is limited as required by giving 
a lower value to one or two of the capacitors, care being taken to 
provide sufficient voltage drive in the preceding stage to allow for the 
increased reactance of the capacitance at low frequencies. 


The source resistance should be as high as possible to overcome the 
effect of the non-linear input resistance, otherwise there will be signi- 
ficant distortion. The source resistance is determined mainly by the 
collector load resistance of the preceding stage. 

The distortion produced by the variation of a' with collector current 

Page 127 


is of lesser importance, provided the collector-current excursion does 
not extend below 0-3mA. This distortion is a minimum for the OC71 
when the current is in the region of 1 to 2mA. 

Peak collector currents of up to 50mA are permissible for the OC70, 
OC71 and OC75, but it is inadvisable to design for peak currents 
of greater than 25mA where low distortion is required. 


In many a.f. stages, the signal level is such that noise need not be 
considered. Provided the volume control is suitably sited, only the input 
transistor need be operated under minimum-noise conditions. Good 
results are obtained with a collector current of 0-3 to 0-5mA and a 
source impedance of 500 to 2000£2. To avoid introducing noise from 
the resistors, high-stability types may have to be used in the first stage. 

All stages prior to the volume control should have a dynamic range 
of at least 10:1. 

If the source impedance is high and no transformer is used, the tran- 
sistor should be operated in grounded emitter at as low a current as 
possible (0-3 to 0-5mA). Silicon transistors, because of their low leakage 
currents, are probably more suitable than germanium transistors in 
such circumstances. 


A high input impedance may be obtained with a grounded-emitter 
stage by adding resistance in series with the input. Alternatively an 
unbypassed resistance R e may be inserted in series with the emitter, 
the input resistance then being approximately 

rb+(r e +Re)(l+a')- 

The first method is preferable as the circuit can be stabilised against 
temperature changes without appreciable loss of gain. 

Grounded-base stages have a very low input impedance, and give 
a high voltage gain when feeding into a high impedance. 

Single-stage Feedback 

Two methods are possible for applying negative feedback to a single 
stage: (a) emitter feedback from an unbypassed emitter resistance; 

Page 128 


(b) shunt feedback from collector to base. 

Distortion and noise are decreased, and the cut-off frequency 
increased, by the feedback factor. Thus 6dB of negative feedback 
(feedback factor of two) halves the distortion and noise and doubles 
the cut-off frequency. The signal-noise ratio, however, is not improved. 

Multistage Feedback 

For a specified overall gain it is preferable, in principle, to include as 
many stages as possible in the feedback loop(s). Because of the wide 
spreads and low cut-off frequencies of most audio transistors, a feed- 
back loop containing more than two stages tends to be unstable, and 
the suppression of the instability decreases the bandwidth. It has been 
found that a two-stage amplifier using current feedback is 'some sort 
of optimum'. 

If the phase shift in the coupling circuits is neglected, a two-stage 
RC-coupled amplifier is inherently stable with any degree of negative 
feedback. If the transistors are directly coupled, a variable negative- 
feedback resistor may be used as the gain control, without any risk of 
instability. With transformer coupling, the transformers should have low 
leakage inductances if appreciable negative feedback is to be applied. 

Amount of Feedback 

The circuit has to be designed for the required minimum of feedback 
to be present with the lowest-gain combination of transistors. A check 
then has to be made that the circuit will remain stable with the highest- 
gain combination of transistors, and where appropriate with a loud- 
speaker load. 

In a two-stage amplifier designed for a minimum of 6dB of feedback, 
the maximum feedback may be about 20dB. 

In output stages, a design for a minimum of 6dB of feedback gives 
sufficient improvement in performance, without making the transfor- 
mers too expensive. 

Individual amplifiers can be designed with greater amounts of feed- 
back. Any particular amplifier can be checked for stability by verifying 
that the application of feedback does not produce undue peaks in the 
very low and very high frequency regions. 

Frequency Correction 

A non-level response, such as is required for pick-up equalisation, for 
example, may be obtained by including frequency-sensitive components 
in the feedback networks. The feedback networks should be between 

Page 129 


well-defined impedances. 


The stability of a multistage amplifier is sometimes impaired by 
coupling between input and output circuits through the supply. 
Instability is avoided by decoupling the earlier stages by an RC net- 
work. The decoupling capacitance is usually lOOfxF. 

A dry battery shows a marked increase in internal impedance near 
the end of its life. This effect has to be borne in mind when considering 
the amount of decoupling required. 


Table 1 gives operating conditions and performance for an OC71 in 
the basic amplifier stage of Fig. 4. Gain and distortion are for a source 

Fig. 4— Typical RC-coupled stage for OC71 

Operating Conditions and Performance for Basic OC71 Stage 

Operating T am b max = 45°C 





lout lout 

for Dtot = 
























impedance equal to the collector load resistance R c of an identical 
preceding stage, and for a load equal to the input impedance of an 
identical following stage. The current flowing in the l-5kQ load resis- 
tance is the output current considered in the distortion and gain measure- 
ments. The performance therefore applies to one OC71 in a series of 

Page 130 


identical stages in cascade, but is not unrepresentative of what happens 
when a number of fairly similar stages are cascaded. 


Simple amplifiers can be constructed by connecting a number of the 
stages given in Table 1 in cascade. 

Apart from simplicity, there is no advantage in cascading identical 
stages. The output is limited to that of the final stage. In the earlier 
stages there is a sacrifice of gain, because the unnecessarily high operat- 
ing current implies a lower collector load resistance. 

In a correctly designed amplifier, therefore, each stage should be 
designed to supply the drive required by the following stage. 

The overall distortion will be less than that obtained by adding 
together the distortions in the individual stages, because there will be 
partial cancellation of second-harmonic distortion. 


A high-gain preamplifier for operation from a 250V supply is shown 
in Fig. 5. The circuit can be rearranged as in Fig. 6 to accord with the 
positive polarity of the supplies in a succeeding valve amplifier. There 
is no need for a separate supply. In Fig. 6 one side of the output is 
earthed, whereas in the more familiar arrangement shown in Fig. 5 
the output is floating. The performance of the circuit is given in Table 2. 
The preamplifier is suitable for low-impedance microphones and 

The supply voltage is reduced to a safe level across the transistor 
by the flow of the quiescent current (~ 0-7mA) through the 330k£2 load 
resistor and 5-6kD emitter resistor. High voltage gain arises from the 
high resistance in the collector-emitter circuit and the high input 
impedance (about 1MO) of the following valve stage, the shunting 
effect of which is small. 

The circuit will work safely with a supply voltage of up to 275V. 
It has been operated successfully from a 100V supply taken from a 
valve amplifier. 

There is no hum or microphony. Noise is not as low as in a valve 
circuit, but is low enough to make the preamplifier acceptable for most 

R2 is returned to the collector instead of to the supply and provides 
a.c. as well as d.c. feedback, the a.c. feedback path being through R2, 
Rl and the source. Rl ensures that a certain amount of a.c. feedback 
is present even with a very low source impedance. 

Page 131 


R5, the emitter resistor, does not contribute a.c. feedback, even 
though it is not bypassed. The input is applied between base and 
emitter and not through R5, which forms part of the d.c. load. 

The a.c. feedback, apart from improving the frequency response 
and reducing distortion, reduces the input and output impedances. If 
the feedback is undesirable, R2 can consist of two equal or nearly 

xo — A/WV 


Fig. 5 — Basic preamplifier for 250V supply 

Fig. 6 — Rearrangement of 250V preamplifier for feeding valve amplifier 

Performance of Preamplifier for 250V Supply 

Operating T am b max = 45°C 

Output voltage 1 -8V 

Input voltage 5-5mV 

Voltage gain 330 

Output impedance ~ 5k& 

Input impedance 200Q 
Freq. resp. (50O source) rel. to lkc/s 

3dB reduction in gain at 15c/s and 13kc/s 

Total-harmonic distortion 0-4% at 0-5V 

Current drain ~ 0-7mA 

equal resistances with their common point bypassed to the emitter 
by a capacitance of about 0-5(j.F. 

Although in both arrangements of the circuit the input terminals 
are floating, no hum should be introduced provided the preamplifier 
is mounted reasonably close to the microphone or pick-up. With the 
circuit of Fig. 5, the input terminals must never be shorted to earth. 
With that of Fig. 6, the current through the microphone or pick-up 
will not be excessive if the input is accidentally shorted to earth. 
High-stability 5% resistors are recommended. The OC71 could be 
used instead of the OC70, but it would be necessary to re-design the 

Page 132 



Class A operation will be discussed only for single-ended output 
stages, since class B is normally preferred for push-pull circuits. 


Fig. 1 represents a transistor in any configuration operating in class A 
from a supply voltage V cc • First of all it will be convenient to con- 
sider an ideal transistor, for which the knee voltage is zero; an ideal 
transformer will also be assumed. Similar relationships to those 
usually given for a valve will then hold. 

The working point is the midpoint of the dynamic collector load 
line. The quiescent current I q has superimposed upon it a signal 

o o 


Fig. 1 — Transistor in any configuration operating in class A 

current which, with transformer coupling, swings the collector current 
between 2 x I q and zero, the signal-current amplitude being 

ic(pk) = Iq. 

At the maximum collector current, the voltage across the transistor 
is zero, the voltage across the transformer being equal and in the 
opposite direction to the supply voltage V cc • On the reverse half cycle, 
the voltage across the transformer is again equal to V ce , but in such 
a direction as to reinforce the supply voltage. The collector voltage 
therefore swings between zero and 2 x V cc , the signal-voltage amplitude 

Vc(pk) = Vcc- ... (1) 

If the reflected load presented by the transformer is Rl , the 

Page 133 


signal-voltage amplitude is 

Vc(pk) = IqXRl. . . . (2) 

From Eqs. 1 and 2, the required load resistance is 

R - Vcc 
Kl = -y- • 

The load resistance may be represented in the usual way by a load line 
drawn across the output characteristic (Fig. 2). The load line will 


Fig. 2 — Idealised load line 

join the points (2 x V cc , 0) and (0, 2 x I q ), the midpoint of the line 
being the working point (V cc , Iq). 

The maximum output power is the r.m.s. value of the voltage 
excursion times the r.m.s. value of the current excursion, thus: 

Poutmax = -£? • -9L = ^V cc Iq ( = $Iq 2 RL> . 

The quiescent power which has to be supplied to the stage, irrespective 
of the output, is equal to the dissipation at the working point: 

Pq — »cclq- 

The maximum theoretical efficiency is therefore P utmax/P a = 50%. 

The collector dissipation has its maximum value of V C c X Iq under 
zero-signal conditions. 


In practice there is always some series resistance Ra. c . in the output 

Page 134 


circuit. For example, in Fig. 3, Ra. c . consists of the resistance of the 
output-transformer primary (R p ) and the unbypassed resistance in 
the emitter. Also, the minimum allowable collector voltage is not quite 
zero, but equal to or slightly greater than the knee voltage Vknee • Then 

Vcc Vknee 



Pout max 



(V cc — VkneeHq 

Iq 2 Rd. 

Drive Requirements 

For a collector-current excursion of I q , the required base-current 
excursion is 

i - Iq - 

The preceding stage must be able to provide this peak drive current. 
The drive voltage required is given by 

V b = IqRe+Vbe- 

The drive requirements should be calculated for a transistor having 
the minimum a' and maximum Vbe • The input impedance is Vb/ib and 
the input power required is 


The operating conditions of the driver stage and the ratio of the 
driver transformer must be chosen with these values in mind. In 

Fig. 3 — Grounded-emitter output stage with d.c. resistance R p + R e 
Fig. 4 — G rounded-col lector output stage 

principle, the impedance at the transformer secondary should be fairly 
high for low distortion; in fact, however, the transformer usually has 

Page 13S 


a step-down ratio in order to reduce the current swing required from 
the driver stage. 

Grounded Collector 

In the grounded-collector output stage shown in Fig. 4, the emitter 
stabilising resistance can be the resistance of the primary of the output 

The drive currents are the same as for grounded emitter, but the 
peak drive voltage is approximately equal to the supply voltage V cc . 
Thus more input power is required. This circuit can be driven from a 
low source impedance. 

Grounded Base 

A transistor in grounded base requires a very high load resistance to 
give useful power gain. This circuit is, therefore, usually not suitable 
for output stages in which the load resistances are low. 


Simple output stages can be designed using the half-supply-voltage 
principle (pp. 62 and 124-5). A single- transistor amplifier based on 


Fig. 5 — 60mW output stage using OC72 operating on half-supply-voltage principle 

this principle is given in Fig. 5. The load is formed directly by a high- 
impedance speaker, the resistance of which drops half the supply 
voltage. As half the d.c. power is dissipated in the load, the maximum 
efficiency (a.c. power/d.c. power) is 25%; however, there is no output 
transformer to introduce further losses. 

The collector current is preset to 33mA at 25 °C by means of RV2. 
The output power is about 60mW at room temperature, but is con- 
siderably reduced at higher temperatures. A high-output crystal pick- 
up, giving about 3V at 1 -2cm/sec, provides sufficient drive through the 

Page 136 


45: 1 transformer for full output from the OC72. Although the output 
is- meagre, an appreciable volume of sound is generated in a speaker 
having a high flux density ( > 10,000 gauss). 

The cost of the complete amplifier is low, even with a high-efficiency 
speaker, as there is no output transformer and only one transistor is 


The output from the OC26 power transistor, which can be about three 
watts in a conventional class A circuit, is sufficient for most car-radio 
receivers. Such receivers may be all-transistorised or 'hybrid' (contain- 
ing both transistors and valves). The valves in the hybrid receivers are 
specially designed to operate from a 14V supply. 

Low-voltage Operation 

A receiver using conventional valves and operating from a low-voltage 
d.c. source, such as a car battery, must necessarily include some device 
(usually a vibrator) to provide an h.t. line of, say, 200V. Although 
valves have been designed to give perfectly satisfactory performance in 
the early stages of a receiver, using low anode and screen-grid voltages, 
it is not practicable to obtain sufficient power from a valve output 
stage at low voltages. The introduction of power transistors has there- 
fore made possible the design of complete low-voltage receivers, 
since the transistor requires only a low operating voltage. 

'Hybrid' Receiver 

The hybrid car-radio receiver has the following advantages: 

(a) the vibrator and its associated components are not required 

(b) filtering is simplified 

(c) the cost is (at present) less than that of an all-transistor design 

(d) the current drain is considerably reduced. 

The performance is at least equal to that of a conventional all-valve 
receiver with a vibrator supply. 

Two car-radio output stages will now be described, one for an OC26 
output transistor driven by an EF98 valve, the other for an OC26 
driven by an OC71. 

Supply Voltage 

Both circuits (Figs. 6 and 7) have been designed for a nominal line 
voltage of 14V. Occasionally the terminal voltage of the car battery 
may rise to 15V or more. Another complication is that a filter is usually 
included in the supply line to suppress interference, and this filter may 

Page 137 


cause a voltage drop of about 0-5 or 1-0V. The actual line voltage 
therefore may only be determined with reference to a practical design. 
It should be noted that a lower line voltage will reduce the maximum 
available output power. Also, the circuit should be changed if necessary 
to ensure (i) a screen-grid voltage of at least 12 -6V and an anode voltage 
of at least 12V for the EF98 in the valve-driver circuit, or (ii) a line 
voltage of 12-6V for the ECH83 and the OC71 in the transistor-driver 


The circuit for an OC26 output stage driven by an EF98 valve is given 
in Fig. 6. The gain of the OC26 is sufficient for it to be driven by the 
EF98 via a 25 : 1 driver transformer. 

Output Stage 

The quiescent current of the OC26 is preset to 600mA by adjusting RV2. 
With a collector load of 21ft, the maximum peak collector current is 
550mA. This gives a maximum output of 3-6W from the transistor 
and 3-3W into the primary of the output transformer, at about 10% 
total-harmonic distortion. With an output transformer as specified, the 
output power into the loudspeaker is about 2-8W. 


Turns ratio 2-5:1 (3Q speaker) 

Primary resistance <0-75Q 

Secondary resistance <0-25Q 

Primary inductance >30mH at 600mA d.c. 


With a 1Q emitter resistor, the total base resistance is 

6-8+0-75 = 7-55Q, 
where the resistance of the driver-transformer secondary is 0-75Q. 
The maximum junction temperature is 90°C at an ambient tempera- 
ture of 55°C, provided the thermal resistance between mounting 
base and ambient does not exceed 2-6°C/W. The transistor under 
these conditions is always thermally stable. 

If this thermal resistance is increased to 3-3°C/W, the maximum 
junction temperature becomes 100°C. (The transistor is rated for a 
maximum of 200 hours intermittent operation at this temperature.) 
The thermal stability, however, is much impaired, as the stability 


is now 0-8. This is dangerously near 10, the limit for stability. It is 

Page 138 


recommended, therefore, that the thermal resistance should not 
exceed 2-8°C/W. 


A minimum of 3dB of local negative feedback is applied to the OC26 
through the collector-base resistor R4. Apart from reducing distortion, 
the feedback also safeguards the transistor against excessive collector 
voltage under overdrive conditions. The feedback of 3dB is sufficient 
for this purpose, because of the small spread in the drive voltage of 
the OC26. 

Driver Stage 

The maximum base current required by an output transistor having the 
minimum a' of 20 is 600/20 = 30mA peak. Local negative feedback 

Fig. 6 — OC26 output stage with EF98 valve driver 

accounts for a further 14mA peak, so that the total drive current 
required is 44mA peak. A drive voltage of 330mV should also be 
available at the secondary of the driver transformer, allowance having 
been made for the flow of base current through the secondary. 

With a 25:1 driver transformer, the EF98 has to provide an 
output-current amplitude of 44/25 — 1 -76mA peak and a voltage of 
330x25 = 8 -25V peak. This drive can be obtained from the driver 
valve at less than 5 % total-harmonic distortion, provided the voltage 
V a k across the valve is equal to or greater than 12V, and provided 
V g 2k is equal to or greater than 12-6V. 

The operating conditions of the EF98 are: 
V a)t = 12V, V g2 k = 14V, Rgi k = 10MQ, R a = 4-4kQ. 

Page 139 



Turns ratio 25 : 1 

Primary resistance <300Q 

Secondary resistance <0-75Cl 

Primary inductance >6H at 6mA d.c. 

The driver transformer should be phased such that an increase in the 
collector current of the OC26 corresponds to a decrease in the anode 
current of the EF98. Although the second-harmonic distortion in the 
output will not be reduced, because the distortion components of the 
transistor and valve do not cancel, the transformer must be phased 
in this way to keep overdrive within safe limits. 

Reduction in Size and Weight 

The circuit shown in Fig. 6 does not represent the most compact 
design which is possible with a valve driver. A thermistor may be in- 
corporated in the bias network, and the quiescent current of the output 
stage reduced. It would then be possible to use a smaller heat sink 
for the output transistor, and to reduce the value, and therefore bulk, 
of the bypass capacitor C2. However, where saving of weight and space 
are important, a better solution is to use a transistor driver. 


The second circuit for a car-radio output stage (Fig. 7) is novel in that 
it incorporates a transistor driver. This circuit has been produced 
primarily with the valve-cum-transistor ('hybrid') receiver in mind. The 
EF98 valve driver is replaced to considerable advantage by an OC71 
transistor, the rest of the receiver being equipped with valves. However, 
the circuit will be equally suitable for all-transistor receivers, provided 
the preceding stages develop the input required by the OC71. 

The usual saving in space and weight is effected when a transistor 
replaces a valve. To follow up this advantage, the heat sink and bypass 
capacitor C7 have also been reduced in size. However, the main advan- 
tage in adopting an OC71 transistor as the driver is an increase in gain 
of approximately 6dB. This increase in gain enables overall negative 
feedback to be applied. The sensitivity has been kept the same as in 
the valve-driver circuit, but the total-harmonic distortion is only about 
3% instead of 10%. 

Also, with a transistor driver, overdrive conditions are easier to 
avoid, because the knee voltage of the driver transistor, and thus the 
maximum available drive voltage, is much more accurately defined 
than for a valve. 

Page 140 


Smaller Heat Sink and Bypass Capacitor 

The area of the heat sink can be reduced by approximately 30 % by 
accepting a slight reduction in output power, and by improving the 
thermal stabilisation at the higher ambient temperatures by means of a 

The quiescent current in the output stage has been reduced from 
600mA to 500mA, thus reducing the output power from the transistor 
from 3-6W to 3-2W. In a practical class A stage, the efficiency is always 
less than 50 %, so that the transistor dissipations at normal room tem- 
perature are 7-8W and 6-8W respectively. 

As the ambient temperature increases, the resistance of the thermistor 
falls and prevents the collector current from rising. Whereas at the 
maximum ambient temperature, the collector dissipation rises to 
about 9W in the valve-driver circuit, in the transistor-driver circuit, 
it remains virtually constant at 6-8W. 

Fig. 7 — OC26 output stage with OC71 driver 

The resistances in the potential divider across the base of the output 
transistor can be increased by a factor of about two, with a similar 
decrease in bypass capacitance. In fact it is possible to decrease the 
bypass capacitance by a factor of four, from 2000[i,F to 500[xF. With 
the extra drive available from the transistor, there is no appreciable 
reduction in output or frequency response. 

A similar reduction in the size of the heat sink, and some reduction 
in the value of the capacitor, could be obtained with the valve-driver 
circuit. However, as the sensitivity is not so great, and less feedback 
is available to improve the frequency response, the capacitance could 
not be reduced by a factor of four. 

Page 141 


Output Stage 

The OC26 has a 05Q. emitter resistor as part of the d.c. stabilisation 
network. This resistor is not bypassed to a.c, so that some local negative 
feedback is applied to the stage. 

The quiescent current is preset to 500mA by means of RV1 1, and the 
maximum peak collector current is approximately 450mA. With a 
collector load of 29Q, the maximum output from the transistor is 3-15W, 
giving 3W into the primary of the output transformer, at about 3 % 
total-harmonic distortion (with overall negative feedback). Using an 
output transformer as specified, the output power into the loudspeaker 
is about 2-6W. 

These figures are derived below by way of a worked example. 

Turns ratio 2-9:1 (3C1 speaker) 

Primary resistance < 1 -0Q 

Secondary resistance <0-3ft 

Primary inductance >45mH at 500mA d.c. 


The variation of current gain with collector current causes the load line 
of a transistor class A output stage to shift with changes in the signal 
conditions. The output also changes and the optimum load may be 
slightly different from that obtained by the simple theoretical deter- 
mination given below. The optimum loading is usually found by 

A maximum output power of 3W is required into the primary of the 
output transformer. (On account of the undecoupled 0-5Q emitter 
resistor, a slightly higher output of 3-15W is required from the tran- 
sistor.) The maximum signal-voltage amplitude available is the collector 
supply voltage of 14V minus the knee voltage of 0-4V, thus 
v C (pk) = 14—0-4 — 13-6V. The total load Rl required for the transistor 
is given by 

Rl = (!5^)?. 29 . 5a 

and the maximum collector-current excursion is therefore 

136 A <n a 
ic(pk) = 2q^ - 450mA. 

As the emitter resistor is part of the total load on the transistor, the 
load required across the primary of the output transformer is 
29-5-0-5 = 29Q. 

Page 142 


The output power into the primary of the transformer is 
(^IV. 29 = 3W. 


The collector loading of 29ft consists of the resistance of the primary 
winding of the output transformer plus the reflected load impedance. 
For a really efficient transformer, the primary resistance should not 
exceed ID. The reflected load is then 29— 1 = 28ft. For a 3ft loud- 
speaker, the resistance of the secondary winding should not exceed 
0-3ft for reasonable efficiency. The total secondary impedance is thus 
3 -3ft, and this should be reflected as 28ft into the primary. The trans- 
former turns ratio is 



„ „ = 2-9:1. 


The power into the primary of the output transformer is 3W. The 

power into the load with the winding resistances as specified above is 

'•0X28+1 X 3+CT3 = 2 ' 6W ' 


The output transistor is stabilised by means of a thermistor placed 
across the lower resistor, R12, of the potential divider which provides 
the base-bias voltage. When the ambient temperature rises, the fall in 
the resistance of the thermistor effectively prevents the collector current 
from increasing. At an ambient temperature of 55°C, the collector 
current is 510mA, which is only 10mA greater than at 25 °C, and the 
maximum dissipation is then 6-8W at the nominal battery voltage of 
14V. The transistor is stable against thermal runaway under the above 
conditions, provided the thermal resistance between the mounting base 
of the transistor and the surroundings is 3-3°C/W. The transistor is 
also stable for a battery voltage of 15V at an ambient temperature 
of 50°C. 

The maximum junction temperature with the specified thermal 
resistance is approximately 85°C, and this is well within the maximum 
of 90°C for continuous operation. Furthermore, the maximum is 
100°C for 200 hours intermittent operation. An amplifier could be 
designed to take advantage of these maximum junction-temperature 
limits. However, the sensitivity would be reduced, as better d.c. stabilisa- 
tion would be required to prevent thermal runaway. Also, a loss in 
output power would occur, unless a high-value and bulky capacitor were 

Page 143 


added to decouple the emitter resistor of the OC26. The circuit of Fig. 7 
is considered to give the best compromise between considerations of 
size and sensitivity. 

Negative Feedback 


The local negative feedback from the unbypassed emitter resistor tends 
to linearise the input impedance of the OC26. This feedback reduces 
the possibility of overdriving output transistors which are characterised 

All dimensions 
rounded to nearest 
even number of mm 

Exploded view of heat sink/chassis for OC26 in laboratory model of 
transistor-driver circuit. 

by a low base-emitter voltage. The feedback increases the drive voltage, 
and not the drive current, required by the output stage. 


The maximum input voltage required at the grid of the triode of the 
ECH83 is basically 80mV r.m.s. for full output. The maximum input 
voltage required if an EF98 is used instead of an OC71 is 150mV r.m.s. 
Nearly 6dB more gain is thus obtained from the transistor driver, and 
this increase in gain is used to add overall negative feedback. 

The feedback loop is taken from the secondary of the output trans- 
former to the cathode of the ECH83 triode. The feedback has to be 
taken to the input of the valve, in order that a higher output current 
will not be required. The grid of the triode is not a suitable point at 
which to apply the feedback, as the input impedance of the stage, 
and thus the load on the preceding diode, would be reduced. This 
effect is particularly impoitant when the gain control RV1 is set to give 

Page 144 


minimum resistance. The 100Q feedback resistor in the cathode circuit 
does not affect the working of the heptode section of the valve, since 
it is decoupled to r.f. by the 0-1 fxF capacitor C4. 

The feedback loop, although it includes two transformers and three 
stages of amplification, can be made stable for considerable amounts 
of feedback. 

Driver Stage 

The maximum base current required by the OC26 for the maximum 
peak collector current is 500/20 = 25mA peak. A total drive voltage of 
550mV should be available at the secondary of the driver transformer; 
this includes the voltage drop across the secondary winding and also 
the voltage across the unbypassed emitter resistor R13. If the voltage 
available across the primary of the driver transformer is 10V, a driver- 
transformer ratio of 10/0-55 = 18:1 is suitable. The current swing 
in the primary is then 25/18 = l-4mA peak, and the minimum direct 
collector current of the driver transistor must be l-9mA, to avoid 
swinging below 0-5mA, where the characteristic is more non-linear and 
there is excessive distortion. 

The triode section of the ECH83 gives 50(zA peak into the lkD load 
presented by the input impedance of the OC71. Total-harmonic 
distortion is less than 5 %, although more current is obtainable with an 
increase in distortion. With a base current of 50(jlA, the driver transistor 
must have a current gain of 1 -4 x 10 3 /50 = 28 in order to drive an 
OC26 having the minimum a' to the full output of 3-2W. For the great 
majority of OC71 transistors, less than 50[xA drive will be required 
from the valve for full output. Occasionally, if both the OC71 and 
OC26 have the minimum a', up to 61\xA peak will be taken from the 
triode. However, as the valve is within the feedback loop, the increase 
in distortion at the output will be small. 

The operating conditions for the ECH83 triode are : 

V b = 12-6V; R a = 4-7kQ; R gi = 10MQ; 

D to t(for Ut = 50(xA) = 5% 


The resistance of R8 in the negative supply line should be such that 
the supply to the driver stage at the nominal battery voltage (14V) 
is not greater than — 12-6V. The collector-emitter voltage is then 
12-6-l-9x0-82= 11V. 

The d.c. voltage rating V c max of the OC71, with the 6-5kQ source 
impedance existing in the circuit, is 12-7V. The collector-emitter 
voltage of the transistor will not exceed this rating, even if the battery 

Page 145 


voltage rises to 16V, provided the collector-emitter voltage is 11 V 
with a 14V battery. 

The a.c. source impedance of the OC71 is 3kQ and the v C ( P k) max 
rating of the OC7 1 under this condition is greater than 25V. Provided 
the supply voltage to the driver stage is not greater than — 12-6V, the 
a.c. rating will not be exceeded. ■*- 


Turns ratio 18:1 

Primary resistance <150Q 

Secondary resistance < 1 Oii 

Primary inductance > 10H at 2mA d.c. 

As in the valve-driver circuit, in order to avoid overdriving of the OC26, 
the phasing of the driver transformer should be such that an increase 
in the collector current of the OC26 corresponds to a decrease in the 
collector current of the OC7 1 . 

Frequency Response 

The gain of the complete audio amplifier (triode valve plus OC71 and 
OC26) falls to 3dB below the mid-frequency value at 70c/s and 8kc/s. 
This response is for nominal transistors, with the overall feedback 
applied, and with a 200kQ source impedance. 


The voltage inputs required at the grid of the triode of the ECH83, 
with the overall negative feedback applied, are given in the following 

Output from OC26 Minimum-a! transistors Nominal transistors 
Full output (3 -0W) 150mV 116mV 

50m W 22m V 17mV 

Although a drive current of 67[xA peak is required at the base of the 
OC71 when both the OC71 and OC26 have the minimum a', for nominal 
transistors, the typical figure is 30fxA peak (without overall feedback). 

Page 146 



In class B push-pull operation, one transistor conducts while the other 
is cut off. Thus the two transistors amplify alternate halves of the 
waveform, and their inputs are 180° out of phase. 

In principle the two transistors should be biased to cut off. Strict 
adherence to this condition, however, results in 'crossover' distortion 


•■? — y- 

Fult sine wave — — — 


' ' / 

Average music — — 

' S' 
















; f 







' ,"' 

>' j'' /' 


J / J 



' ' 'J 



> // 




f / 

' / 


Fig. 1 — Direct current plotted against nominal peak output power, for average music 
and for full sine-wave drive. 

which is unpleasant to the listener. This distortion is overcome by 
applying a small forward bias to each transistor. 

The main advantages of class B push-pull as compared with class A 
operation are 

(a) low quiescent current, and 

(b) high efficiency at full output. 

Because the current drain is low, class B operation is favoured for 
the output stage for equipment operating from dry batteries. 

The battery drain depends on the signal being handled. The average 
current consumption on music is about a third of that on maximum 
sine-wave output. 

In Fig. 1, the full lines show the direct current with average music 

Pag* 147 


plotted against the nominal peak output power, for various battery 
voltages. The broken lines are for full sine-wave drive. 

The output from the class B push-pull pair is about five times the 
collector dissipation of a single transistor. The maximum theoretical 
efficiency (a.c. power/d.c. power) is 78-5%, and usually in practice an 
efficiency of 70 to 75% can be realised. 

The transistors could be connected in common base, common 
emitter or common collector. Common-base operation is not suitable 

Fig. 2 — Basic common-emitter circuit 

Fig. 3 — Common-emitter circuit with integral secondary resistances 

for practical circuits because of its low power gain. Common-emitter 
and common-collector circuits, and also a split-load arrangement which 
lies between the two, are described in this chapter. 


The basic common-emitter circuit is shown in Fig. 2. The potential 
divider RV1, R2 provides the quiescent bias necessary to eliminate 
crossover distortion. The sum of the quiescent currents of the two 
transistors is preset to the design value by means of RV1. An emitter 
resistor R e is required for thermal stability. R2 and the resistances of 
the transformer secondaries also influence thermal stability. In general 
the secondaries should have as low a resistance as possible. 

Occasionally it will be cheaper to use transformers having high 
secondary resistances. These resistances can form the lower part of the 
bias potential dividers, as shown in Fig. 3. The quiescent current of 
each transistor is preset to the design value by means of RV1 and RV2. 

In Fig. 4, the centre-tap of the output transformer is connected to 
the supply. The transistors still operate in the common-emitter mode, 

Page 148 


even though the collectors are taken directly to the negative supply. 
This circuit has the advantage, for power transistors, that both tran- 
sistors can be mounted directly (without mica washers) on a common 
heat sink, which is connected to the negative supply terminal. 

Whenever it is possible to adopt a centre-tapped supply, the single- 
ended push-pull circuit of Fig. 5 offers the least costly solution. A 
high-impedance speaker provides the load, instead of a conventional 
low-impedance speaker and output transformer. The performance is 


|r., < 

J R ei 



Jr V2 



? r> 







Fig. 4 — Common-emitter circuit with collectors connected to negative side of supply 
Fig. 5 — Transformerless output stage for centre-tapped supply 

the same as for the other circuits, if each half of the centre-tapped 
battery has the same voltage as that of the supply in the other circuits. 

RV1 and RV2 in all the above circuits can be fixed instead of variable 
for transistors with close Vbe spreads. The bias conditions given in the 
published data should be adopted. 

Thermal Stability 

None of the operating conditions should be changed without investigat- 
ing the thermal stability. The quiescent currents must be set to the 
design values, the stability being impaired at higher settings. 

The maximum dissipation and the thermal stability depend on the 
total thermal resistance of the transistor from junction to ambient. 
With power transistors, the thermal resistance largely depends on the 

Page 149 


heat sink and the manner in which the transistor is mounted. The 
thermal resistance (junction to ambient) must always be that required 
by the design or lower. 

Normally, users of this book will be concerned with operating or 
servicing existing equipment which may include a heat sink, or with 
building up designs which include a specification for a suitable heat 
sink. Provided the function of the heat sink is understood, and nothing 
is done to impair its efficiency, there should be no trouble from thermal 
instability. Further information on heat sinks will be found in Chapter 9. 

Transformer Turns Ratios 

The turns ratios of the output and driver transformers can be calculated 
simply from the information supplied. The usual equation applies to 
the output transformer, namely 

Re = Rp+m2R s 
where R c is the load impedance per transistor, R p is the resistance of 
the primary, R s is the total resistance of the secondary and speaker, 
and the turns ratio is m+m:l. In an efficient transformer, R p will 
be not more than 5 % of R c , and the resistance of the secondary 
winding will be not more than 5 % of the speaker resistance. 

The published data show the peak drive voltage required at the 
secondary of the driver transformers. Let this be \2. Let the peak 
voltage available at the transformer primary be vi. Then the turns 
ratio n:l is given by n = V2/V1. 


In the split-load circuit (Fig. 6) part of the load is in the emitter. This 
arrangement reduces crossover distortion at lower battery voltages 
and lower ambient temperatures. 

In conventional class B common-emitter output stages, a quiescent 
bias voltage is applied, which results in a linear transfer characteristic 
at nominal battery voltage. As the battery voltage decreases, the 
quiescent bias decreases, and the transfer characteristic becomes non- 
linear, resulting in crossover distortion. 

In the split-load circuit, the presence of part of the load in the emitter 
means that a higher drive voltage Vb is required at the base for the same 
peak collector current i C (pk) . The non-linear region of the characteristic 
forms a smaller proportion of the peak drive than before, and the 
distortion is reduced. 

Best results would be obtained with the entire load in the emitter 
(that is, as in common-collector amplifiers) but the loss in gain may 
be too high. The fraction of the load in the emitter circuit is thus a 

Page ISO 


compromise between gain and performance. As a rough guide, the 
tolerable variation in battery voltage increases by the same factor as 
the increase in base input voltage. 

The base input voltage is greater than that of an equivalent common- 
emitter amplifier by the voltage developed across the emitter load, 
and the driver and earlier stages must be designed accordingly. 

Crossover distortion at low battery voltages (or low ambient tem- 
peratures) is lower than for a common-emitter amplifier. In all other 
respects the performance of the two types of circuit is similar. 


The transistors can also be used in common-collector class B push-pull. 
The drive current is the same as for a similar common-emitter circuit, 
but the drive voltage is approximately equal to the sum of the output 

Fig. 6 — Split-load circuit 

voltage and the drive voltage for common-emitter operation. Therefore 
much greater drive power is required. The driver transformer must 
have a step-up ratio, and the inductances of the driver and output 
transformers must be higher than for a common-emitter output stage. 

Crossover distortion and non-linearity distortion are less than in 
common-emitter or split-load circuits, but the gain is considerably 
reduced. Common-collector stages are therefore adopted only in 
special circumstances. 


The design procedure of common-emitter class B stages will now be 

Page 151 


described. A complete specimen design will not be given, but some of 
the points will be illustrated by figures based on a 15W output stage. 
The procedure can be modified to cover the design of various types of 
output stage, as indicated in the preceding sections. 

In designing an output stage, it is necessary to determine: (a) the 
maximum dissipation of the transistor; (b) the peak currents and 
voltages; (c) the optimum load; (d) the output power; and (e) the input 
current and input voltage. The distortion at various outputs can also 
be evaluated. 

Allowance has to be made for the full spread of the transistor 
characteristics. All of (a) to (d) above are very nearly independent of 
transistor characteristics, provided that when calculating (e), the input 
current and voltage requirements, the extreme-limit characteristics 
are used. 

Calculation for Zero Knee Voltage 

The design equations are best illustrated by giving them for transistors 
having zero knee voltage, and then making adjustments for the departure 
from this ideal. In practice the power output is about 85 to 95 % of the 
ideal. The following relations can be proved for a pair of transistors 
having zero knee voltage: 

Maximum peak voltage = supply voltage V cc 
Maximum peak current = =r^ » 


where R ce is the collector-emitter load per transistor 
Maximum output power of the pair (P ou t max) = ^ 
Efficiency at maximum power = 78 -5 % 

2 (Vcc) 2 

Maximum dissipation per transistor = 

2 2R ce 

~ £ Poutmax. 
A graph showing the dissipation and efficiency related to peak signal 
is given in Fig. 7 for sine-wave drive. Maximum dissipation occurs 
when the actual maximum collector-current amplitude is 0-638 of the 
ideal, and the efficiency at this current is 50%. Consider the load line 
shown in Fig. 8. Then: 

Maximum output voltage (ideal) = 14V peak 

Maximum output current (ideal) = 3-3A peak 

Load per transistor R ce = 14/3-3 = 4-250 

Maximum output power from the pair (ideal) = £(14)(3 -3) = 23 • 1 W 

Maximum dissipation per transistor (ideal) = K23-1) = 4-6W. 

Page 152 


Effect of Knee Voltage 

In the above expressions, the minimum collector-emitter voltage is 
assumed to be zero. In fact, the minimum collector-emitter voltage at 
a particular collector current is the knee voltage given in the published 
data. Because of the non-linearity in the vicinity of the knee voltage, 
a slightly higher voltage is taken as the minimum for the purpose of 
design. This minimum voltage is proportional to the collector current 
and can be represented by a straight line on the output characteristic 

£ o* 





02 04 0-6 0-8 10 

, Peak signal current or voltage 
Max peak current or voltage 

Fig. 7 — Efficiency and relative dissipation plotted against relative signal 

(Fig. 8). Tan <j> is the minimum voltage for a particular current I c divided 
by the current I c , and has the dimensions of a resistance. This method 
of dealing with the minimum collector-emitter voltage is convenient, 
as tan § is the same for all collector currents. 

The effect of the minimum voltage is to decrease the maximum peak 
output voltage. Therefore the maximum peak current, the maximum 
output power, and the efficiency at maximum output are all slightly 
decreased. For example, in Fig. 8, tan <j> == 0-4O. With V cc = 14V and 
R ce = 4-250 as before; 

Maximum output voltage = V cc 



R ce + tan 4> 
Vcc =3 0Apeak 

= 12 -8 V peak 

Maximum output current = ^— 

R ce + tan . 

Maximum output power = J(12-8)(3 0) = 19-2W. 

Page 153 


In Figs. 7 and 8, 

i c max. actual v c max. actual 


i c max. ideal 

v c max. ideal 


= 0-915, 

tan <|> 

so that 

Efficiency at maximum output power = 72-5%. 

Since the maximum dissipation occurs at 0-638 of I c max, the currents 
and voltages at which maximum dissipation occurs are unaffected by 








K | 

1 1* 


' ' 


Fig. 8 — Output characteristic and load line 

the knee voltage. Therefore the maximum dissipation is the same as for 
zero knee voltage. 

Emitter Resistor 

The emitter resistor is in the collector-emitter loop and forms part of 
the load on the collector. As the emitter resistor cannot be decoupled, 
this resistor introduces a loss of output power. 

If R c is the reflected useful load in the collector, and R C e is the total 
load in the collector-emitter loop, 

Rce == Re+Rc • 
Page 154 




Useful output power = P ou t max . 5— • 

For the same example as before: 

Reflected useful load = 4-25-0-5 = 3-750 

Useful output power = 19-2. j^7 

= 16-9W. 

Quiescent Current and Power Dissipation 

In general the maximum value of the quiescent current is small in 
comparison with the maximum collector current. It can be shown that 
the effect of the quiescent current on the maximum collector dissipation 
is to increase the dissipation by s(IqV cc ). 

In an output stage giving 15W output, the maximum quiescent 
current might be 130mA at an ambient temperature of 55 °C. The in- 
crease in dissipation would then be 

0-2x0-13x14 = 0-36W. 
Hence the maximum collector dissipation is 

4-6+0-36 = 5W. 

The quiescent current does not affect the maximum output power. 

Drive Current 

The maximum drive current can be found from ic/a', where i c is the 
maximum collector current required (e.g. 3-0A) and the value of a' is 
the minimum which may occur for the type of transistor in question. 

Drive Voltage 

The drive voltage required to produce a collector current i c consists 

(a) the change in the voltage across the emitter resistor, equal to 

ieRe — IqRe 

(b) the change required in the base-emitter voltage, allowance being 
made for transistor spreads. The required change in base-emitter 
voltage is equal to the base-emitter voltage at the collector current 
i c ( = Vbe(ic)) minus the base-emitter voltage at the collector 
current I q ( = V be (iQ)) 

(c) the additional voltage required for the base current ib to flow 
in the resistance of the base circuit, that is, the circuit consisting 
of the transformer secondary and the equivalent bias source 

Page 155 


The voltage required in the base circuit is ibRb , where Rb is the total 
base-circuit resistance. 

Therefore the drive voltage required is 

(ie— Iq)Re + Vbe(i c )— Vbe(I q ) + ibRb • 

(The above considerations of drive current and voltage include the 
assumption that maximum Vbe occurs with transistors on the low-gain 
limit. The actual combination of characteristics will not always be as 
unfavourable as this, and the maximum drive required is then some- 
what less.) 

The drive power required (r.m.s.) is given by 
Kidrive peak x Vdrive peak). 
This value of drive power includes the loss in the driver-transformer 


The main causes of distortion are non-linearity of the I c /Ib (transfer) 
characteristic, crossover distortion, and mismatch of the two transistors. 


The a! of a transistor decreases with increasing current over most of 
the current range. The a'/Ic characteristic is, however, controlled so 
that the maximum variation in gain is restricted to within acceptable 
limits. The effect of this type of non-linearity is to produce predomin- 
antly third-harmonic distortion. 

The distortion can be calculated for any given or derived I c /Ib 
characteristic. The following method is simple and sufficiently accurate. 
Let i c i be the peak collector current at which the distortion is to be 
calculated. Then: 

(a) from the characteristics obtain the base current (represented by 
ibi) which corresponds to the collector current i c i 

(b) determine the collector current (represented by i C 2) which 
corresponds to |ibi 

(c) calculate the third-harmonic distortion from 

D3 = i-HrTr 100 %- 

lcl/lc2+ 1 


The difference in a' of the two transistors causes the two halves of a 
sine wave to be unequal (assuming current drive). The distortion at 
any output current can be calculated for any two specified a'. The 

Page 156 


second-harmonic component can be obtained readily from the follow- 
ing table, where D2 is correlated with the ratio of the two a'. 
Ratio of a' 1-1 1-2 1-3 1-4 1-5 1-6 1-7 1-8 1-9 20 

D 2 % 2-4 4-6 6-5 8-3 9-8 11-5 13-0 14-5 15-5 16-5 

If the maximum ratio of the two S! is 1-35:1, the corresponding value 
of D 2 is 7-5%. 

The total-harmonic distortion (neglecting harmonics higher than the 
third) is given by 

D t ot= V{(D 2 ) 2 +(D 3 ) 2 }. 


Any difference in the input impedances of the two halves of the circuit 
causes second-harmonic distortion. The amount of distortion is 
dependent also on the source impedance, high source impedances 
giving less distortion. In general, with a transistor driver stage, the 
source impedance is high, and mismatch of the input impedances is not 
of any serious consequence. 


If the cut-off frequencies of the transistors are not equal, the phase 
shifts of the two halves of a sine wave are unequal, the difference becom- 
ing larger as the frequency is increased. The resulting distortion is 
somewhat similar to crossover distortion and is unpleasant, the inter- 
modulation distortion of frequencies higher than the cut-off frequencies 
being fairly high. In designing complete amplifiers, it is therefore 
preferable to limit the upper cut-off frequency of the amplifier to below 
the cut-off frequency of the output stage, the frequency-restricting 
circuit being inserted prior to the output stage. 


Crossover distortion occurs if the change-over in current from one 
transistor to the other is not smooth. If the composite transfer charac- 
teristic is not a straight line, but shows discontinuities or changes in 
slope, then crossover distortion results. This distortion results in a large 
amount of intermodulation distortion, and is the most objectionable 
form of distortion occurring in class B output stages. 

To eliminate crossover distortion, it is necessary to bias the transistors 
such that the resultant transfer characteristic is a straight line. The 
optimum bias conditions, as given in the published data, ensure that 
the performance is satisfactory under all normal operating conditions. 
Crossover distortion will re-appear, however, under extremes of 
temperature or battery voltage. 

Temperature-sensitive elements, such as n.t.c. thermistors or junction 

Page 157 


diodes, can be used to compensate for variations in ambient temperature ; 
and non-linear resistive elements, such as copper-oxide rectifiers, or 
lamps of suitable resistance, can give compensation for voltage varia- 

Emitter feedback offers one way of minimising crossover distortion 
at low battery voltages. This method was discussed in the section on 
split-load stages (p. 150). 


Hole storage in the base region produces small pulses of current at 
the instants when the current in the two transistors changes over. 
These pulses may make the driver transformer ring, if its leakage 
inductance is high. The use of bifilar windings or a CR damping circuit 
across the primary will decrease this effect, where it is troublesome. 


Some practical class B push-pull output stages are given in Figs. 9 to 
13. The values of the dropper resistors are not given, as these will 
depend on the current drawn by the preceding stages. Complete ampli- 
fiers based on the circuitry given in Figs. 11 and 13 are included in 
Chapter 16 (pp. 171-2). 

The driver and output transistors for the circuits illustrated are 
supplied together in a plastic packet. The LCR3 packet contains an 
OC82D driver transistor and a matched pair of OC26 output transistors. 
The LFH3 packet consists of an OC81D driver transistor and a matched 
pair of OC81 output transistors. 

Fig. 9 — Common-emitter 7W push-pull amplifier for hybrid car-radio receiver. 
The maximum drive (r.m.s.) required at the base of the driver tran- 
sistor is 70(i.A and 40mV for full output. The output transistors must 
be mounted on heat sinks giving a thermal resistance of 7°C/W per 
transistor. Ambient temperature not to exceed 55°C. 

Page 158 


The advantage of these packets is that the possible spread in per- 
formance is reduced and circuit design is simplified. Transistors from 
one packet should not be interchanged with those from another, or 
loss of performance may result. 

Transformers for all the LCR3 and LFH3 circuits can be supplied 
by R. F. Gilson Ltd. For the 25Q. speaker for Fig. 12, application may 
be made to TSL. 

Fig. 10 — Common-emitter 500mW push-pull amplifier. The input current 
(r.m.s.) for full output is 6-4 to 10[zA. For operation at ambient 
temperatures not exceeding 45°C, the transistors may be mounted 
in free air, without cooling clips. 

Fig. 11 — Split-load 540mW push-pull amplifier. The input current (r.m.s.) 
for full output is 5-5 to 17-5(/.A. For operation at ambient temperatures 
not exceeding 45°C, the transistors may be mounted in free air, 
without cooling clips. 

Page 159 


AH resistors ± 10% except 
where otherwise shown 

Fig. 12 — Single-ended 500mW push-pull amplifier. The input current (r.m.s.) 
for full output is 10 to 15-5|i.A. For operation at ambient temperatures 
not exceeding 45°C, the transistors may be mounted in free air, 
without cooling clips. 


Fig. 13 — Common-emitter 1W push-pull amplifier. The input current (r.m.s.) 
for full output is 16-5 to 26-5{xA. The transistors must be mounted 
on heat sinks of 5x7cm of 16 s.w.g. aluminium. The transistor is 
bolted down to its heat sink by means of a close-fitting cooling clip 
(obtainable from Kimber- Allen Ltd. or distributors).Tamb ^ 45°C. 

Page 160 



The OC57, OC58, OC59 and OC60 form a range of transistors specially 
developed for hearing aids. These transistors are cylindrical in shape 
and are only 4mm long and 3mm in diameter. They are already available 
to the industry and will be generally available in due course. These 
types are in keeping with the general trend towards miniaturisation 
of the components of the hearing aid. 

Using these transistors, a complete hearing aid can be mounted in 
spectacle frames, or as a clip-on unit concealed behind the ear or in 
the hair or clothing. 

Three hearing-aid circuits will now be described as an illustration of 
the general design principles of audio amplifiers, and in particular for 
their bearing on the operation of transistors at low levels and from low 
supply voltages. These circuits are not intended to be typical of modern 
commercial practice. The first two circuits are RC coupled ; they repre- 
sent a line of development which has led to extremely compact designs, 
such as the third circuit, which is directly coupled. 


Before special transistors were introduced for the purpose, hearing aids 
were designed using the standard small-signal a.f. transistors, that is, 
the OC70 and OC71. Fig. 1 shows a circuit of this type, in which the 
supply voltage is 2-4V. 

The usual considerations of d.c. stability show that better stabilisa- 
tion is required at lower collector supply voltages. Good stability is 
therefore provided in the circuit of Fig. 1, but this necessarily means 
using a large number of resistors and capacitors. Also, the OC70 and 
OC71, which are approximately cylindrical in shape, are about 5mm in 
diameter and 15mm long. Consequently the complete hearing aid is 
bulky by modern standards, although of course a tremendous advance 
over designs equipped with subminiature valves, as regards both size 
and battery consumption. 

Later on, some reduction in size was made possible by substituting, 
in this circuit, the OC65 for the OC70 and the OC66 for the OC71. 
The OC65 and OC66 were in a metal construction, which was about 

Page 161 


3mm x 4mm in cross-section and 7mm long. These two transistors were 
the forerunners of the OC57 series. 

Demoded though this circuit is, it illustrates many of the techniques 
discussed in earlier chapters, and therefore will be discussed more fully 
than its present commercial position would warrant. 


The first two stages are operated at currents of only 0-3mA to minimise 
noise. The gain control RV6 is sited between the first and second stages 

Insert earpiece 
Ikflat lOOOc/s 
250*1 d.c. 

Resistors ±5% 
Capacitors 6vd.c. wkg. 

Fig. 1 — Four-transistor RC-coupled hearing aid 

to ensure a low noise level, while keeping this control clear of the 
feedback loop. 

Input Stage 

The input impedance of the amplifier is about lkQ. This matches the 
impedance of the electromagnetic microphone, which is specified as 
IkO at lOOOc/s. An arrangement similar to that described previously 
for the 250V high-gain preamplifier provides a.c. and d.c. feedback, 
except that extra resistance is not required in series with the microphone. 
The unbypassed emitter resistor R4 does not contribute a.c. feedback, 
since the input is applied between base and emitter. 

Page 162 


Driver Stage 

The collector current in the driver transistor Tr3 is 0-5mA. A low 
collector load resistance is required in this stage to ensure adequate 
swing up to full drive without clipping. 

Output Stage 

The output stage is designed round an OC71 to provide an output of 
nearly 2mW to the earpiece. Since in class A the maximum theoretical 
efficiency is 50%, a collector dissipation of about 4mW is required. 
This is provided by a collector current of 2mA and a collector-emitter 
voltage of 2V, the load impedance being 2V/2mA = lkQ (at lOOOc/s). 
The d.c. resistance of the earpiece is 250Q, which at 2mA drops 
2 X250 = 0-5V, and leaves nearly 2V across the transistor. 

The collector current is set to 2mA by the choice of the feedback 
resistor R13, the value of which should be a' times the lkQ load 
impedance. Thus for a transistor having an a' of 47, R13 should be 

R13, in addition to biasing the transistor, provides a.c. and d.c. 
feedback, the d.c. stability being satisfactory with this method of 
stabilisation because of the higher collector current of 2mA, and 
because R13 is chosen to match individual transistors. 

Feedback Loop 

In addition to the a.c. feedback provided by R2 in the input stage and 
R13 in the output stage, the gain is stabilised by 12dB of negative 
feedback taken over the last three stages. A feedback voltage propor- 
tional to the output current is taken from across R14 in the emitter 
of the output stage, and injected into the second stage in series with 
the bypassed emitter resistor R8. The 20. resistance for R14 can be 
made from approximately 7in. of Eureka wire 0092in. in diameter 
and having a resistance of 10-6ft/yard (1 in. = 2-54cm, 1 yd. = 36 in.). 


The power gains in the four stages are 17, 21, 18 and 30dB, giving a 
basic total of 86dB. As there is 12dB of loop feedback, the overall 
power gain reduces to 74dB. 

The frequency response of the amplifier itself is 2dB down at 150c/s 
relative to the response at middle frequencies, the electrical response 
being virtually flat from lOOOc/s to 7kc/s. The overall acoustical 
response of a complete hearing aid would depend primarily on the 
microphone and earphone. 

The performance is satisfactory up to an ambient temperature of 
40°C (104°F). 

Total-harmonic distortion, measured at a test frequency of 400c/s, 

Page 163 


is 5% at full output. 
Current drain is 3 -5mA with the correctly matched value for R13. 


The use of the latest hearing-aid transistors in an RC-coupled circuit 
is illustrated by Fig. 2. 

In this circuit the electrical power gain is about 85dB using only 
three transistors. The earpiece should have an impedance of 650Q. and 
a d.c. resistance of 200O, and the microphone should be a magnetic 
type of 2kQ impedance. The loss in the microphone and earpiece is 
about 35dB, so that the acoustical or air-to-air gain is 85—35 = 50dB. 

In the OC58 output stage, the collector current is about 2mA, and at 
full supply voltage the output power is then about 0-5mW, which is 

Fig. 2 — Three-transistor RC-coupled hearing aid 

sufficient for this application. R7 does not have to be set for individual 

In the OC57 driver stage and OC59 input stage the collector current 
is 0-25mA. Rl and R4 provide sufficiently good d.c. stability to permit 
a wide range of ambient temperature. R4 introduces some a.c. feedback, 
but there is little loss in gain, because the input impedance falls and 
reduces the loss in the coupling network. CI prevents a.c. feedback 
in the input stage. 

By siting the volume control between the input and driver stages, 
the amplification of contact noise is reduced. 

With a mercury cell (such as the Mallory RM625) the decoupling 
components for the first stage, R3 and C3, become unnecessary. 

Average current drain is 2 -7mA, and a life of about 100 hours can 
be obtained from a Mallory cell RM625. Total-harmonic distortion is 
5% at 0-4mW, and the electrical frequency response is flat to within 
0-25dB from lOOc/s to 4kc/s. 

Page 164 



The previous circuit, if it is used with a mercury cell so that R3 and 
C3 may be omitted, still contains five fixed resistors and three capacitors. 
The need for a subminiature hearing aid which can be concealed in a 
pair of spectacles or a hair slide has led to the consideration of direct 
coupling for the amplifier. Such a circuit is shown in Fig. 3, and apart 
from the usual microphone, earpiece, battery and volume control, 

> 3-9 

> kfi 



HI ,z=6oon 

(— L-l lot Ikc/s 

® L ^® L d^) 

R3(-R f ) 


Fig. 3 — Three-transistor directly coupled hearing aid 

the only components required are the three transistors, three resistors 
and one capacitor. 

The circuit is experimental rather than a production prototype. 
A pre-requisite of any large-scale production of a hearing aid utilising 
this circuit would be an examination of the effects of the production 
spreads of the transistors. 

Choice of Circuit 

The only directly coupled circuit which will give a useful power gain 
with three transistors consists of three grounded-emitter stages in 
cascade. Stabilisation is provided by a d.c. negative-feedback loop. 
This feedback loop governs the choice of circuit. Overall feedback is 
applied from output to input, so that the d.c. gain of the whole ampli- 
fier is used to compensate the temperature-dependence of the transistor 

The volume control is sited so as not appreciably to affect the d.c. 
working conditions. 

Page 165 


Operating Conditions 

To achieve low power consumption, the first two stages are operated 
at 0-3mA, which is the lowest practicable collector current. The 
collector voltage in each of these stages is limited to the base-emitter 
voltage of the following transistor. The collector voltage, which will 
be below the knee of the typical output characteristic, is therefore 
about 120 and 170mV for Trl and Tr2. 

From the supply voltage (1-3V) and the resistance of the ear- 
piece, a working point of V c = 1 -02V, I c = 1 -6mA can be derived for 
the output stage. This is the ideal working point at which maximum 
output can be obtained without clipping. In practice, to secure a useful 
performance over a wide range of ambient temperature, the collector 
current of the output stage should be set to 2 -2mA at an ambient 
temperature of 25°C. The collector voltage is then 910mV. The ideal 
working point is only realised at lower ambient temperatures. 

Decoupling Resistance 

The value of the decoupling capacitor Ct ( = CI) is determined by the 
loss of gain which can be tolerated in the working-frequency range. 
A loss of 6dB at lkc/s is considered permissible. A time-constant 
Cf X Rf of about one second is required, and as Rf , which is chosen 
to suit the characteristics of the output transistor, may be as low as 
160kQ, a nominal capacitance of 6fxF is specified for Cf . 


The nominal air-to-air gain is estimated to be about 48dB. The per- 
formance should be acceptable over a range of ambient temperature 
from to 39°C (32°F to 102°F). The predicted electrical response 
is within l-5dB of the response level at lkc/s for the range 300 to 

The current drain is 2- 8mA at 25 °C, giving a life of about 90 hours 
from a Mallory cell, type RM625. 

Page 166 



Output transistors fall into two classes : the larger, higher-power ones 
in a construction similar to that of the OC26; and the smaller, lower- 
power types in a construction similar to that of the OC72. Output 
stages for both groups of transistor are designed along the lines given 
in Chapter 14. In the second group, matched pairs of transistors 
operating in class B give output powers at various levels up to the 
region of 1-OW. Such transistors are already a familiar feature of 
portable radios and record players. 

Complete amplifiers using matched pairs of OC72 are not untypical 
of low-power audio amplifiers, and two 200mW circuits based on the 
OC72 are described in some detail in this chapter. Two circuits are 
then given for 540mW and 1W amplifiers using the LFH3 (OC81D 
driver plus a matched pair of OC81 output transistors). An experimental 
high-quality amplifier which provides an output of about 5W and uses 
OC22 and OC42 transistors is included at the end of the chapter. 


The operating conditions quoted in the published data for the OC72 
show that a matched pair of transistors can give an output of 390mW 
when the transistors are fitted with the specified cooling fin, and 275mW 
without the cooling fins. 

The circuits about to be described are the familiar ones for 200mW; 
these have two advantages for the general user: (a) cooling fins and 
special stabilisation arrangements are not required; (b) the trans- 
formers are available from a number of manufacturers. 

Two versions of the 200mW audio amplifier are described, one for a 
4-5V and one for a 6V supply. The 6V circuit is treated as the basic 
version. The performance of the two circuits is compared in Table 1. 

Fig. 1 shows the 6V version of the amplifier, which gives the full 
200mW output for an input of about 400m V. This input may be pro- 
vided by a crystal pick-up connected to the high-impedance input 
terminals XX. 

A low-impedance input may be connected to YY, an input of 3mV 
then being sufficient for full output. 

The amplifier is suitable for a portable record player or, with a 
front-end, for a portable receiver. The same 6V supply may be used 
to feed the amplifier and the turntable motor. 

Page 167 


Circuit Description 

If the output stage is compared with the recommended biasing con- 
ditions given in the published data, it will be seen that some compro- 
mise of d.c. stability has been accepted. The emitter resistor has been 
discarded and, because of the expense, a thermistor has not been 
adopted in the biasing network. The stability is sufficient, however, to 
permit operation up to an ambient temperature of 45°C (113°F), 
without an undue increase in crossover distortion. 

RV11 is a preset control which should be adjusted to give a com- 
bined quiescent current of 1 -2mA in the output stage. 










All resistors should be ±57. 

Fig. 1 — 200mW audio amplifier for supply voltage of 6V 

Feedback from the secondary of the output transformer is applied 
directly to the collector of the input stage via R13. The value of R13 
must be chosen to suit the impedance of the speaker, using the table 
in the top left-hand corner of the circuit diagram. 

The OC71 driver stage is designed round a collector current of 
1 -5mA. Thermal stability is ensured by the lkQ emitter resistor. 

The supply line is decoupled by R10-C4, thus reducing the distortion 
which would otherwise occur with the increase in battery impedance 
during life. The collector of the driver being much less sensitive to 
the feedback than the base, it has been possible to incorporate the filter 
at a low-current point, between the collector and base connections to 
the supply. 

The low collector current in the input stage gives a comparatively 
high input impedance of several kQ's at the terminals YY. An input 
impedance of about 500kO is needed, however, to match a crystal 
pick-up, and this is achieved at the terminals XX by inserting a 330kQ 

Page 168 


resistor Rl in series with the 1MQ gain control RV2. The load on the 
pick-up will be greater than 1MQ when the control is turned well down, 
so that maximum bass output is obtained at low volumes. When only 
the high-impedance input XX is to be used, CI may be reduced from 
10 to 0-1 (jiF. 

Resistors of 5 % tolerance are recommended, and coupling capacitors 
of 10(xF prevent any serious reduction in bass response. The emitter 
resistors in the input and driver stages are bypassed by 100>F capaci- 
tors. In addition, the filter capacitor C4, to be effective at all audio 
frequencies, has to be 100(xF. The capacitors should be 6V d.c. working. 


With a matched pair of transistors operating in push-pull, there is no 
d.c. component of magnetic field, and each half of the primary must 
be capable of carrying the peak current of 83mA without saturation. 
As it is possible for the a' of the two transistors to be mismatched by 
up to 35 %, the output transformer should be constructed to allow for 
unbalanced direct currents. 

The whole of the primary should have an inductance of 0-5H, and 
its resistance should be as low as possible ( < 3Q). The secondary 
resistance should be less than 5% of the load resistance. Leakage 
inductance should be as low as possible. Turns ratios (whole primary 
to secondary) for various speaker loads are given in the table on the 
circuit diagram. The output transformer is rated at 250m W. 


The primary inductance should be 10H at a primary direct current of 
l-5mA. Lower primary inductances give poorer bass response. The 
turns ratio is 3-5:1 + 1. The d.c. resistance of the primary should 
be less than 200Q and that of each half of the secondary less than 
50Q., and the d.c. resistances of the two halves of the secondary should 
be equal. Leakage inductance should be as low as possible. The second- 
ary should preferably be bifilar wound to give closer coupling between 
the two halves. The driver transformer is rated at 2m W. 


A version of the 200m W amplifier for operation from a 4-5V supply 
is shown in Fig. 2. Like the 6V version, this amplifier can be driven to 
full output by a crystal pick-up connected to the terminals XX. The 
two circuits are similar in principle, but there are many differences in 
detail. Attention will be drawn only to the most important of these. 

The output stage is biased by presetting RV1 1 to give a total quiescent 
current in the stage of l-3mA±10% at 20°C (68°F) or l-6mA±10% 

Page 169 


at 25°C (77°F). With this adjustment, the circuit will operate satisfac- 
torily at ambient temperatures from 15°C (59°F) to 45°C (113°F). The 










7 3<1 



4-6 <1 












R6 < Jm 



All resistors should be ±57. 

Fig. 2 — 200m W audio amplifier for supply voltage of 4-5V 


Performance of 6V and 4-5V 200mW Amplifiers 

Output power 
Total-harmonic distortion 









at XX 



Frequency response 
3dB fall relative to lkc/s at: 

50c/s & 8kc/s 

35c/s & 8kc/s 

Current drain 
zero drive 
average music 
sine-wave drive, max output 

5 to 6mA 
12 to 15mA 


range of operating temperature may be extended down to 10°C (50°F), 
by substituting for R12 ( = 180O) a resistor of 270Q shunted by a 
Varite thermistor VA1039. 

Different values for the feedback resistance and turns ratio (whole 
primary to secondary) of the output transformer are specified, as 
shown in the table in the top left-hand corner of the circuit diagram. 
Each half of the primary should have a d.c. resistance of less than 2-50. 
The driver stage is designed for a collector current of 3mA and the 
turns ratio of the driver transformer is 2 : 1 + 1. The primary inductance 
of this transformer should be 10H at a primary direct current of 3mA, 
and the d.c. resistance of the primary should be less than 150Q. The 
resistance of each half of the secondary should be less than 750. 

Page 170 



Other possible operating conditions will be found in the published data 
for the 2-OC72. These conditions apply to stages which differ from those 
just described in that they contain an emitter stabilising resistor (Fig. 3) 
and have a higher quiescent current. Most of the conditions permit 

Fig. 3 — Modified output stage with shared emitter resistor 
higher output power, all of them give lower crossover distortion. 
However, if these conditions are used as the basis of design, the sensi- 
tivity will be worse and the driver stage and driver transformer will 
have to be redesigned, in addition to using a different output transformer. 


The 1W amplifier (Fig. 4) uses the LFH3 transistors in the circuit 
of Fig. 13, Chapter 14 (p.160). The response, relative to that at l-5kc/s, 

Fig. A — 1W amplifier. Colne transformers 06005 (T1) and 06006 (T2) 

is 3dB down at 85c/s and 6kc/s. The OC81D and OC81 transistors must 
be mounted on heat sinks of 5 x7cm of 16 s.w.g. aluminium for opera- 
tion at ambient temperatures up to but not exceeding 45°C. The tran- 

Page 171 


sistor is bolted down to the heat sink by means of a close-fitting cooling 
clip (obtainable from Kimber-Allen Ltd. or distributors). 

The 540mW amplifier (Fig. 5) uses LFH3 transistors in a split-load 
circuit similar to that of Fig. 11, Chapter 14 (p. 159). The bias network 
has been slightly modified to simplify the construction of the driver 

l,C3,C4,C5 are 6V d.c. wkg. C2 is 12V d.c. wkg 

Resistor tolerances +10% unless otherwise stated 

Fig. 5— 540mW amplifier. Colne transformers 06003 (T1) and 06004 (T2) 

transformer. The response, relative to that at l-5kc/s, is 3dB down at 
HOc/s and 4-3kc/s. If the LFH3 transistors are mounted in free air 
without cooling clips, ambient temperatures of up to 45 °C are permis- 

The sensitivity in both circuits is 250 to 350mV for full output 
at 10% total-harmonic distortion and for a 9V supply. 

Circuit Description 

An experimental high-quality amplifier is illustrated in Fig. 6. The 
output power is 5W into a resistive load, the total power gain being 
63dB at Ikc/s. To extend the frequency response at high and low audio 
frequencies, high-frequency transistors and direct interstage coupling 
are used throughout. A single negative-feedback loop further extends 
the frequency response and reduces internal non-linearity distortion. 
The amplifier will deliver its rated output power at ambient temperatures 
up to 45°C. 

The output transistors are operated in class A push-pull, because 
low distortion is a requirement. OC22 transistors have been chosen 

Page 172 


because of their high cut-off frequency and the good linearity of the 
Vb/Ic characteristic. The transistors do not need to be exactly matched, 
but they should not be grossly mismatched. The maximum permissible 
collector-junction temperature of 90°C allows a quiescent collector 
current of 600mA at an ambient temperature of 45°C with a 14V supply. 
Temperature stabilisation is effected by the emitter resistance R20 

All resistors jw±l07. unless otherwise stated 

Electrolytic capacitors 12V d.c. wkg. 

Fig. 6 — Experimental 5W high-quality audio amplifier 

which, being common to both emitters, introduces no degenerative 
feedback in a class A stage. This resistor may be made up from two 
3-3Q, 3W wirewound resistors in parallel. 

Each OC22 is driven by an OC42 'emitter follower', the base voltage 
of the driver stage, and therefore the base bias of the output stage, 
being derived from a resistive potential divider. Two potential dividers 
couple the collector and emitter of a conventional phase splitter (Tr2) 
to the bases of the two driver transistors. Equal a.c. (signal) loads are 
presented to the collector and emitter of Tr2. 

A conventional grounded-emitter amplifier (Trl) drives the phase 
splitter. Negative feedback from the output transformer is applied to 
the emitter circuit of this stage. 

The quiescent current in each OC22, and therefore the balance, 
is determined by the settings of RV8 and RV13. The following 
sequence of adjustments for the initial setting-up of these controls is 
recommended : 

Page 173 


(1) with suitable meters in the collector circuits of Tr5 and Tr6, 
RV8 is adjusted until the current in Tr5 is 600mA 

(2) RV13 is then adjusted to make the current in Tr6 equal to 600mA 

(3) RV8 is re-adjusted to return the current in Tr5 to 600mA 

(4) repeat (2) and (3) in that order as necessary. 


The performance of an experimental model of the amplifier, with 

17dB of negative feedback, was: 

Max. power output at lkc/s 5W (into 3Q load) 

Frequency response 

(a) 1W = OdB -3dB at 3c/s & 50kc/s* 

(b) 5W = OdB -3dB at 7c/s & 40kc/s 

Total-harmonic distortion 

(a) lWat lkc/s 0-3% 

(b) 5Wat lkc/s 0-6% 

Input impedance . . . . 8-3kfi 

Sensitivity 16-8(xA for 5W at lkc/s 

*A peak of +2-5dB occurs at 40kc/s following a continuous increase from 
lOkc/s (OdB) 


The output transformer should have a turns ratio of 1-65 + 1-65:1 
(for a 3Q load). The primary resistance should be less than 0-5Q 
(each half) and the secondary resistance less than 0-1Q, and the primary 
inductance should be at least 1H. C5 may be inserted if necessary, 
depending on the output transformer used. This capacitor was not 
incorporated in the experimental model. 

Heat Sink 

Each OC22 must be provided with a heat sink having a thermal 
resistance of 2°C/W. 

The heat sink for each OC22 may consist, as in the experimental 
model, of a 9|-in. strip ( ~ 24cm) of extruded aluminium, type Noral 
6182. The strip as supplied is 4x 1£ in. ( ~ 10 X 3cm). The finish may 
be bright and the strip may be mounted in any position. 

An equivalent heat sink has been calculated for 14 s.w.g. Duralumin 
sheet metal. Each heat sink should be 8x8in. (20-5x20 -5cm). The 
finish again may be bright, but the heat sink must be mounted vertically, 
although it may be folded along the vertical axis if required. 

Whichever of the two types of heat sink is adopted, each OC22 
should be insulated by a mica washer and insulating bushes. A thin 
smear of silicone grease should be provided between the washer and 
the heat sink. The two heat sinks allow a large margin of safety. 

Page 174 



In this chapter two versions of a public-address amplifier are described, 
one for a 14V and one for a 28V supply. The output power is 15W at 
less than 4 % total-harmonic distortion. The sensitivity is sufficient for 
the amplifiers to be driven fully by low-impedance microphones, and 
the performance is more than adequate for the intended application. 

The 14V version of the circuit can be regarded as the basic amplifier 
and will be described first; it is then only necessary to describe the more 
important differences in the 28V circuit. 

The performance of the two versions of the circuit is compared in 
Table 1. The amplifiers are suitable for operation at ambient tempera- 
tures which normally do not exceed 45°C, but occasional rises up to 
55°C are permissible. 


The basic amplifier (Fig. 1) is designed for a supply of 14V, which is 
the average voltage of a fully charged 12V accumulator. 

An output stage of the type shown in Fig. 4 of Chapter 14 has been 
adopted. The output is provided by two matched OC26 operating in 
the common-emitter configuration in symmetrical class B push-pull. 

The output stage is preceded by an OC26 driver and by OC72 and 
OC71 amplifier stages. 

Output Stage 

Although in the circuit of Fig. 1 the collectors are connected to the 
negative fine, the transistors operate as common-emitter amplifiers. 
The advantage of this arrangement, it will be remembered, is that the 
output transistors can be mounted directly on a common heat sink 
connected to the negative fine. 

Emitter resistances of 0-5Q. are necessary for thermal stability at the 
higher ambient temperatures. To minimise crossover distortion, the 
quiescent currents are set individually to 30mA at normal ambient 
temperatures ( ~ 25°C) by means of RV19 and RV20. Each of these 
variable resistors forms the upper half of a potential divider biasing 
the appropriate transistor. The lower half is formed by the resistance 
of half the secondary. 

Page 175 


c rWH-i 

L - / \A/V\-yMM&Z 

MEflfiffltr-AAA/V- 1 




ID 1 









— VWv- 










Page 176 


The optimum load per transistor is 4-25Q, of which 0-5Q is provided 
by the emitter resistor and 3-75Q has to be matched to the speaker. 
A centre-tapped choke, which effectively acts as a 2:1 auto trans- 
former, provides the 4:1 impedance ratio for matching the 15Q 
speaker. This choke is cheaper and more efficient than a conventional 

The peak current on full drive is 3-OA, and the maximum transistor 
dissipation is 5W at an ambient temperature of 55°C. The maximum 

All dimensions rounded 
to nearest even 
number of 

Fig- 2 — Approximate dimensions of heat sinks for OC26 output transistors made 
from 2mm blackened-aluminium sheet metal. 

junction temperature is 85°C, with the transistors mounted on a heat 
sink giving a total thermal resistance from junction to ambient of 
6°C/W. The transistors are thermally stable at this junction temperature. 
Compact and cheap heat sinks can easily be made to give the required 
total thermal resistance (Fig. 2). 

Negative feedback is applied from the output to the base of the driver 
to minimise distortion. 


For a 15Q speaker a centre-tapped choke is suitable. 

Total d.c. resistance < 0-2Q. 

Total inductance > lOOmH 

Driver Stage 

The driver stage consists of a single OC26 in a conventional class A 

Page 177 


circuit. The collector current is adjusted to 125mA by means of RV14. 
The collector dissipation is considerably less than in the output stage, 
and the total thermal resistance, from junction to mounting base, only 
has to be less than 15°C/W. This value is achieved very easily by 
mounting the transistor with a mica washer on the chassis. 


Turns ratio 2-5:1 + 1 (bifilar secondary) 

Primary inductance > 500mH at 125mA d.c. 

Primary d.c. resistance < 60. 

Secondary d.c. resistance 5ft +50 

The resistance of each half of the secondary should be 5Q±10%. If 
the resistance of the windings is less than this, external resistances must 
be added to make up the required value. 

Amplifying Stages 

The first stage is equipped with an OC71 and amplifies the signal 
from the microphone input. This stage is followed by 0C71 and 
OC72 current-amplifying stages. The circuit is conventional except 
that the second OC71 is directly coupled to the OC72, and a.c. and d.c. 
negative feedback is applied over these two stages. This method of 
coupling requires fewer components and provides better temperature 
stability than two conventional RC-coupled stages. 

Crossover distortion in the output stage increases at higher fre- 
quencies. It is therefore preferable to limit the upper cut-off frequency 
of the amplifier to about 7kc/s. C6 in the feedback loop provides the 
necessary limiting. 

Reproduction of frequencies below 150c/s is not desirable in public- 
address systems. The smaller value used for CIO (4(xF instead of 10yF) 
provides a convenient method of limiting the low-frequency response. 


Fig. 3 shows the version of the circuit for operation from a 28 V supply 
(usually two fully charged 12V accumulators in series). This circuit is 
of the transformerless push-pull type shown in Fig. 5 of Chapter 14. 
The 3-75Q load impedance is provided directly by the speaker. There 
are also some differences in component values and ratings between 
the two versions of the circuit. 

If a centre-tapped 28V supply is not available, an artificial centre-tap 
can be provided. A possible method is to connect two 24Q, 10W 
resistors in series across the supply, with the centre-tap decoupled by a 
1000(xF capacitor of 50V d.c. wkg. 

Page 178 



Page 179 



Turns ratio 3-8:1 + 1 (birilar secondary) 

Primary inductance > 600mH at 80mA d.c. 

Primary d.c. resistance < 8ft 
Secondary d.c. resistance 5ft+5ft 

The resistance of each half winding of the secondary should be made up 
if necessary to 5ft ±10%. 

Performance of 14V and 28V Circuits 

14V 28V 

Current Consumption I q 220mA av. 150mA 

speech and music ~ 800mA ~ 400mA 

Sensitivity (for full output) -2 fxA • 1 \lA 

impedance lkQ 0-2mV 0-lmV 

Distortion (at full output) < 4% < 4% 
Frequency Response Flat within 3dB from 150c/s to 7kc/s 


A higher output power, of perhaps 20W, will be available from OC26 
circuits, although the sensitivity will necessarily be reduced. While 
such circuits will be similar in principle to those just described, the 
transformers and almost all the component values will be different. 

Page 180 



An i.f. amplifier can be designed for a.m. reception whose performance 
compares favourably with that obtainable with thermionic valves. 
This is perhaps the most interesting example of the method to be fol- 
lowed when using transistors as h.f. amplifiers. Equations for calculating 
the component values are given in fuller treatments of the subject. 


Internal feedback is of major importance at high frequencies. It can 
produce instability in much the same way as feedback through the 
anode-grid capacitance of a triode valve. Even if oscillation does 
not occur, the bandpass characteristic may be highly asymmetrical. 

Internal feedback can be neutralised by means of external feedback. 
The design of the amplifier then becomes relatively straightforward. 

When both real and imaginary parts of the feedback are cancelled, 
the process is known as unilateralisation. 


A fall in gain with frequency is experienced with all transistors. The 
cut-off frequency is the point at which the current amplification factor 
falls to 3dB below its low-frequency value, and for an h.f. transistor 
is at least several Mc/s in grounded base. This loss of gain sets the 
limit to the usefulness of a transistor at high frequencies. 

The gain in a grounded-emitter circuit falls off more rapidly with 
frequency than in grounded base, so the cut-off frequency needs to 
be well above the frequency of operation. 

For an OC45 the maximum theoretical gain is about 38dB at 470kc/s. 
The stage gain in a practical narrow-band amplifier would normally 
be from 2 to 12dB lower than this, because of the insertion loss of the 
coupling elements, or intentional mismatching losses. 


Two requirements therefore emerge for a transistor required to operate 
at h.f. First, the internal feedback should be small and not subject to 
too great a spread. Second, the transistor will be suitable if the cut-off 
frequency f a in grounded base is about ten times the operating frequency. 

Page 181 

470kc/s I.F. AMPLIFIER 

The OC45 has been designed specially for these requirements, the 
average value of f a being 6Mc/s. 


Fig. 1 shows the circuit of a 470kc/s amplifier, the first stage of which 
will be taken as a design example. The d.c. conditions are set up in 
the normal way. The emitter current is 1mA, and with an emitter 
resistor R4 of 680Q the voltage at the emitter is about — 0-68V. The 
voltage between base and emitter (across the emitter junction) is small, 

Fig. 1 — Complete circuit of i.f. amplifier for 470kc/s 

the voltage at the base being about — 0-85V. The base voltage is fixed 
by the potential divider formed by Rl ( = 56kO) and R2 ( = 8-2kQ), 
and by the voltage on the a.g.c. fine. 

CI and C2 are bypass capacitors and C4 is the tuning capacitor for 
the coil. Unilateralisation is provided by R3 and C3 which form a 
feedback path from output to input. The tap on the coil allows a more 
convenient value for C4, as will be explained later. 

Tl and T2 can be identical and are designed for a transistor output 
impedance of 28kQ and an input impedance of 800Q. 


Setting up the a.c. conditions requires a detailed knowledge of the 
characteristics of the transistor, and these are best expressed by means 
of an equivalent circuit. 

There are a number of equivalent circuits which give an accurate 
representation of transistor characteristics. Any one of these could be 

Page 182 

470kc/s I.F. AMPLIFIER 

used as a basis for designing an h.f. amplifier without affecting the 
final result, and in fact the various equivalent circuits are only re- 
arrangements of each other which can be obtained by normal circuit 

The most convenient equivalent circuit for grounded-emitter i.f. 
stages, however, is that shown in Fig. 2. The amplification of the 


Cl3£OOjtrnho) b ' 

bo VWV - 

r bb' 

Fig. 2 — Equivalent circuit of OC45 at any frequency 

transistor is represented by a current generator acting directly across 
the output terminals. The value of the current generator (in milliamps) 
depends on the signal voltage Vb'e between the points b' and e. The 
current generator is therefore conveniently designated by a mutual 
conductance, g m , in mA/V. For most purposes, where the operating 
frequency is well below f a , the values of all the circuit elements 
(including the generator) can be regarded as independent of frequency. 

This circuit is called the hybrid 7t equivalent circuit. The values 
on the diagram are for an OC45 operating at V ce = — 6V, I e = 1mA. 

Provided only one frequency is being considered, the much simpler 
form shown in Fig. 3 can be used. (This simplified arrangement is the 
normal 7c equivalent circuit.) All the circuit elements except C3 depend 

-^WAA, II — 

7550n 9 . 9 | pF 


Fig. 3 — Equivalent circuit of OC45 at 470kc/s 

on the operating frequency. The only connection between the input 
and output is the feedthrough path of R3 = 7-55kQ in series with 
C 3 = 9-95pF. 

The value of the current generator has to be changed, since it must 
now be defined in terms of the true input voltage Vbe instead of the 
voltage Vb'e between b' and e. 

Page 183 

470kc/s I.F. AMPLIFIER 

The difference between these voltages is caused by the resistance 
r b b' between the points b and b'; this resistance is an internal resistance 
(hence the small r) in series with the base, and is contributed by the 
base material lying between the active region of the base layer and the 
base contact. 

Because of the voltage drop in rbb' , the mutual conductance is 
reduced to 35mA/V, this value being denoted by G m to distinguish it 
from the g m = 38mA/V of the previous circuit. 


The effect of the internal feedback is that a voltage produced at the 
output terminals by the current generator will produce an unwanted 
voltage at the input terminals across the source impedance. 

Neutralisation is effected by connecting a phase-changing trans- 
former across the output (Fig. 4) and incorporating a suitable imped- 
ance in the feedback path. The current fed back to the input through 

Rf Cf 

-AAAA, 1| 

l-26kA 59-7pF 

-A/WV — IK 



7-55kn g.95pF 
I R2> C2! 


Fig. 4 — tv equivalent circuit with unilateralising components added 

the external feedback path is equal in amplitude, but opposite in phase, 
to that fed back through the internal feedback path. So the total 
feedback is zero. 

The external feedback is most effective when the transistor is uni- 
lateralised, that is, when both the real and imaginary parts of the 
internal feedback are cancelled. Thus if the transformer ratio were 
1:1, Rf would be made equal to R3 ( = 7-55kQ) and Cf equal to C3 
( = 9-95pF). In fact a step-down ratio of n:l is used to match the 
high output impedance into the low input impedance of the following 
stage, and Cf = nC3 and Rf = R3A1. For a transformer ratio of 6: 1, 
the values will be R f = l-26kQ and Cf = 59-7pF. 

When the transistor is unilateralised, changes on the output side 
cannot affect the input circuit, and conversely. The transistor reduces 
to a box having independent input and output circuits (Fig. 5). In 
this condition the external feedback elements Rf and Cf merely exert 
a shunting effect on the input and output impedances. This will be 

Page 184 

470kc/s I.F. AMPLIFIER 

taken into account shortly. In practice, the power lost in Ri is often 
negligible and Cf may have only a slight influence on tuning. 


The input and output impedances of the unilateralised transistor can 
be calculated using equations derived from the equivalent circuit. The 




870 \Cj 

PF 'y 


■c ut 

Fig. 5- 

-Equivalent circuit of unilateralised OC45 at 470kc/s excluding effects of 
neutralising components. 

effect of the elements in the external feedback path will not be taken 
into account yet. 

The input resistance R in is calculated as 800Q and this is in parallel 
with an input capacitance cm of 870pF. 

To obtain the output impedance the assumption is made in the first 
instance that the transformer is perfectly lossless. The output resistance 
Rout is then 29kQ and the output capacitance Cout is 38pF. 


The condition for the power lost in Rf to be a minimum is that 

(29,000 , 

/Rout /^ 

11 ~ V^in ~~ V 


where Rm is the input resistance of the following stage. 

The impedance in the external feedback path must therefore be one- 
sixth of that originally calculated, that is, 

R f = — = 1 -26kQ 


C f = 6x9-95 = 59-7pF. 

The shunting effect of the neutralising components on the input 
and output impedances can now be calculated; first Rf and Cf are 
converted into an equivalent parallel combination of resistance and 
capacitance (R p = 26-8kO and C p = 57pF) which shunts the input 
impedance (Fig. 6). 

Because the turns ratio of the transformer is 6: 1, R p and C p appear 
on the output side of the transformer as n 2 R p = 36x26-8 = 965kQ 
and C p /n 2 = 57/36 =j l-6pF. 

Page 185 


The combined input and output impedances are shown in Fig. 7, 
Rout becoming about 28kQ. 

In the final design the nearest preferred values are used for Rf 
( = l-2kn) and C, ( = 56pF). 


After unilateralisation the transistor can be considered as a box with 



Fig. 6 — Equivalent circuit of unilateralised OC45 at 470kc/s including effects of 
neutralising components. 

independent input and output impedances, and the design of the i.f. 
amplifier is straightforward. 

The possible stage gain can now be calculated. To obtain maximum 
power from the box, the load resistance Rl is made equal to the 
output resistance R ou t ( = 28k£2) as in Fig. 8. An inductance must 
also be connected across the output terminals to tune out the output 

Fig. 7 — Simplified equivalent circuit of unilateralised OC45 at 470kc/s including 
neutralising components. 

capacitance. For the time being the inductance can be regarded as 
lossless, that is, of infinite Q. 

Since the current flowing from the current generator will be divided 
equally between the output and load resistances, the power in the load, 
Pout , is given by : 

Pout = [I 2 R] 

= GG m V ln ) 2 RL 
= KG m Vin) 2 Rout • 

The input power is given by 

(Vin) 2 



Page 186 

470kc/s I.F. AMPLIFIER 

Thus the maximum theoretical power gain is : 



KGm) 2 RinRout 

1 \1000/ 
= 38dB. 

777.28,000 = 6700 


The output capacitance of the transistor ( = 40pF) and the input 

i<=out <Rout (°S GmV| 
40 < 28 

> R| - 



Fig. 8 — Unilateralised transistor with load resistance and tuning inductance 

capacitance of the following stage reflected back through the trans- 
former ( = 927/n 2 = 26pF) contribute a total of 66pF to the tuning 
capacitance. However, using 66pF as the sole tuning capacitance would 
give far too wide a bandwidth. The bandwidth can be reduced without 

-WW 1| 

Fig. 9 — Practical arrangement based on circuit shown in Fig. 8: d.c. components 

not shown. 

loss of gain by adding extra capacitance and reducing the inductance 
to tune again to 470kc/s. 

To avoid inconveniently large values for the extra capacitance C 
a tapped primary is used (Fig. 9). The capacitance Or appears as a 
much larger capacitance at the transistor collector, the positioning of 
the tap being determined from 

m_ IC_ 

n ~~ vCt' 

Page 187 

470kc/s I.F. AMPLIFIER 

At the same time the ratio n:l is maintained between primary and 


With an unloaded Q of 100 a sufficient margin of stability will normally 
be assured in a two-stage i.f. amplifier. 

Coil Loss 

The loss in a practical coil depends on the initial Q and is given by 
the following expression: 

Pout with practical coil /Qo— Qw\ 2 

oil = / Qo-Qw V 
>il I Qo J 

Pout with perfect coil 

where Q is the unloaded Q and Q w is the working Q. 
The insertion loss in decibels is 

201o8 qtV b 

and using practical values of Q = 100 and Q w = 56 the loss is 7dB. 


The actual stage gain is the maximum theoretical gain minus the coil 
loss, that is, 38—7 = 31dB. 

The amplifier gain is limited to 31dB by allowing a coil loss of 7dB. 
By not using the maximum gain of the amplifier, the circuit can be 
made stable, allowance being made for the spread in transistor gain 
and other circuit tolerances, such as those introduced by the neutralisa- 
tion components and the Q's of the coils. 

Page 188 



The six-transistor receiver described in this chapter is intended primarily 
for portable equipment, and has a performance comparable with that 
of a conventional four-valve portable receiver. However, the perform- 
ance is also acceptable for normal indoor listening. 

The receiver can be constructed to be carried as easily as a handbag. 
Only slight variations in the design are needed to adapt the circuit for 
either a miniature personal receiver or else a larger transportable 'set 
about the house'. 

The set is designed for medium- and long- wave reception. It com- 
pares favourably as regards sensitivity and output power with typical 
four-valve battery receivers, and has a much lower battery consump- 
tion. The receiver gives 200m W at full output, and has a sensitivity 
of approximately 500{xV/metre (or 20(j.V at the base of the mixer) for 
an output of 50m W. 

The circuit in this as in portable valve equipment is necessarily a 
compromise between performance and cost. Component and tran- 
sistor tolerances have been examined, so that the permitted spreads 
should not create any difficulty. 

Special precautions in the layout are required only to ensure that 
the aerial is not heavily damped by adjacent components, and that 
appreciable feedback does not occur between the i.f. or a.f. stages and 
the aerial. 


Six transistors and one germanium diode are used in the basic form of 
the receiver shown in Fig. 1. The OC44 is the frequency changer and 
the two OC45 form the i.f. amplifier. The detector is the OA70 
germanium diode. Three transistors make up the audio stages, an OC71 
being used to drive a matched pair of OC72 (the 2-OC72) in the 
transformerless push-pull output stage. 

A superhet circuit is chosen for the same reasons that apply to valve 
receivers. The aerial is a Ferroxcube rod, which gives the required 
selectivity and sensitivity and can be coupled conveniently to the 
frequency changer. 

The standard i.f. of 470kc/s is used, and the local-oscillator frequency 
is above the signal frequency, in accordance with normal practice. 

Page 189 















1 ]_l 


t 1 1 





q— ^if^j l_, t 

Page 190 


The OC44 operates as a self-oscillating mixer. The r.f. signals from 
the aerial coupling coil are fed into the base of the OC44, which pro- 
duces its own local oscillation by means of feedback from the collector 
to the emitter. Tracking of the aerial and oscillator coil is obtained in 
a conventional way by means of a tuning capacitor with shaped oscil- 
lator vanes. The i.f. is selected at the collector of the OC44 by the 
first i.f. transformer T3. 

The i.f. amplifier consists of two OC45 operating in unilateralised 
grounded-emitter circuits. The choice of bandwidth is a compromise 
between quality and selectivity. To obtain satisfactory adjacent-channel 
rejection, the i.f. bandwidth has been reduced. The resulting treble 
attenuation will not generally be noticeable during normal listening. 
In miniature receivers, in which the bass response is also reduced, the 
compromise treble response is completely adequate. 

Double-tuned i.f. transformers could be used to improve either or 
both the frequency response and adjacent-channel rejection. 

The third i.f. transformer T5 is connected to an OA70 detector diode 
which provides an audio output and a d.c. output. The d.c. output is 
fed back to control the operating current of the first i.f. transistor, so 
providing automatic gain control. 

The a.f. output from the OA70 is taken to an OC71 driver stage. 
The OC71 is transformer-coupled to a class B output stage consisting 
of the 2-OC72. The output stage is of the transformerless or 'single- 
ended' type, in which no output transformer is required and the loud- 
speaker forms a direct load for the output transistors. A loudspeaker 
with a 35Q. speech coil provides the correct load for an output of 
200m W. Negative feedback is applied to the emitter of the OC71 from 
the loudspeaker terminal. 

A battery voltage of 9V is selected and this voltage is allocated in 
the following way: 

(a) the h.f. transistors work at the collector-emitter voltage of 6V 
which gives maximum gain 

(b) a voltage drop of about IV is allowed across the emitter resistors 
of the h.f. transistors for stabilisation of the working point 

(c) a drop of 2V is allowed across the decoupling resistor from the 
audio output stage, so that the decoupling resistance can be 
high enough to make a very high decoupling capacitance 

Page 191 



In general, in order to obtain the maximum possible power from the 
Ferroxcube rod aerial, the rod should be as long as is practicable in a 
given cabinet. Increasing the diameter of the rod increases the Q and 
the power output of the aerial, but the choice of diameter is limited 
by the increased weight and cost of the thicker rods. The FX1268 
Ferroxcube rod chosen for this receiver is 7in. long and has a diameter 
of approximately fin. 

The aerial is illustrated in Fig. 2. There are separate windings for the 
medium- and long-wave bands. The coils are placed at opposite ends 
of the rod, the centre of each being approximately lfin. from the end. 
This spacing minimises interaction between the coils. Small adjustments 

Fig. 2 — Dimensions of aerial components (1 in. = 2-54cm) 

Coil Details 

Medium-wave coil (single-layer windings) 

Coil AB: 64 turns, 19/00028 bunched conductors 

Coil CD: 6 turns, 19/00028 bunched conductors 

Long-wave coil (wave-wound) 

Coil EF: 41 turns 00076 in. rayon-covered, enamelled wire 

Coil GH: 175 turns 0-0076 in. rayon-covered, enamelled wire 

to the coil inductance can be made when aligning the receiver by 
sliding the coils along the rod. 

If the long- wave aerial coil is left open-circuited during medium-wave 
operation, the coil can resonate with its self-capacitance at a frequency 
in the medium-wave band. The resonance causes heavy damping of 
the medium-wave coil at that frequency. The long-wave coil is there- 
fore short-circuited by SA1 during operation in the medium- wave band. 
The medium-wave coil is left open-circuited during long- wave operation. 

The aerial is coupled to the frequency changer by low-impedance 
coils placed adjacent to the aerial coils. The number of turns on the 
medium-wave coupling coil has been adjusted to reduce the Q of the 
aerial from an initial unloaded value of 200 (measured at lMc/s) to a 
working value of 110 when the aerial coil is loaded by the input resis- 
tance of the frequency changer. The ratio of unloaded to loaded Q of 
almost 2 : 1 gives approximately maximum transfer of power from the 
aerial to the frequency changer, and also effects a reasonable com- 
promise between bandwidth (9kc/s) and second-channel rejection. For 

Page 192 


long-wave operation, the Q of the aerial is reduced from an unloaded 
value of 80 to a working value of 22, again to obtain a bandwidth of 


The frequency changer in a transistor receiver may be one of two types. 
It may consist either of a separately excited mixer requiring a second 
transistor as a local oscillator, or of a self-oscillating mixer in which one 
transistor combines the functions of mixer and oscillator. There is little 
difference in performance with either type, but the self-oscillating mixer 
has two particular advantages, lower cost and better frequency stability. 
Both circuits can be designed for satisfactory frequency stability, 
however, so that cost is the main consideration. 

In this receiver the mixer is of the self-oscillating type and is designed 
round an OC44. The input signal is taken from the aerial coupling 
coil by way of the wave-band selector switch to the base of the tran- 
sistor, the base being the more sensitive input electrode. Oscillator 
feedback is taken from the collector to the emitter through low- 
impedance coupling windings on the oscillator coil. 

The oscillator tuned circuit is similar to that used in valve receivers, 
since the capacitance reflected from the transistor is very small, only 
about lpF. The Q of the oscillator coil is somewhat higher than in 
valve receivers to allow for transistor damping. 

To ensure easy starting for the oscillator, the transistor is biased 
initially in class A by the normal d.c. stabilisation circuit. As the 
amplitude of oscillation increases, rectification of the oscillator voltage 
at the emitter causes a steady negative voltage to be developed across 
the emitter resistor and bypass capacitor. This voltage tends to drive 
the transistor into class B and also stabilises the amplitude of oscillation. 
At the same time the quiescent current increases slightly. 

Operating Current 

The direct emitter current in the OC44 was chosen to be 0-25mA (it 
rises to about 0-3mA when the oscillator is functioning) for two 
reasons : 

(a) the noise level is a minimum when the current is in the region 
of 0-25mA 

(b) the cut-off frequency of the mutual conductance at this current 
is approximately equal to the alpha cut-off frequency f a , which 
is 15Mc/s. 

To maintain a high working Q and good frequency stability, the 

Page 193 


oscillator coil is only lightly loaded by the input resistance of the tran- 
sistor at its emitter. The emitter is thus voltage- rather than current- 
driven at the frequency of oscillation, and so it is the mutual conduct- 
ance, rather than the current amplification factor a, that determines 
the high-frequency performance of the oscillator. At high direct cur- 
rents, the cut-off frequency of the mutual conductance may be con- 
siderably lower than f a . However, if the current is so chosen that the 
internal emitter resistance becomes equal to the internal base resistance, 
the two cut-off frequencies become approximately equal. 

Fig. 3 — Self-oscillating mixer with stray capacitance in tuning capacitor 

A cut-off frequency of 15Mc/s ensures that the internal phase shift 
and fall in gain are small up to the maximum oscillator frequency of 
2-07Mc/s. Furthermore, since the nominal effect of the drop in the 
h.f. performance of the transistor is small, the effect of spreads on the 
h.f. performance is also small. 

Oscillation is maintained in this design at all frequencies under the 
most adverse conditions, that is, when the battery voltage has fallen 
by half, f a is at the lower limit of its permitted tolerance range, and 
rbb' has its maximum permitted value. Also, the oscillator remains 
stable with a transistor which has the maximum permissible f a and the 
minimum rbb' • Squegging can only be made to occur by doubling the 
emitter bypass capacitance. 

The design, therefore, will accept any OC44. 

Page 194 


Tuning Capacitor 

Correct tracking could be obtained with either a conventional padder 
capacitor or a shaped oscillator section, as has been adopted in this 

The value of tuning capacitance is not critical, but must be sufficient 
to provide the desired frequency coverage. The aerial section has a 
capacitance of 175pF, and the oscillator section a capacitance of 123pF. 

If there is no screen between the two sections of the tuning capacitor, 
a stray capacitance exists between them. A self-oscillating mixer which 
includes stray capacitance is shown in Fig. 3. If the aerial is matched 
correctly to the input impedance of the OC44, the stray capacitance 
can cause spurious oscillations at the high-frequency end of the medium- 
wave band. 

The circuit elements in Fig. 3 which control these unwanted oscilla- 
tions are the aerial and oscillator tuned circuits and the stray capaci- 
tance. The circuit resembles that of a triode valve oscillator of the 
tuned-anode tuned-grid type. The unwanted oscillatory voltage appear- 
ing at the collector of the OC44 is stepped up in the oscillator trans- 
former by a factor of 6-55 ( = 72/11), and a corresponding feedback 
current flows through the stray capacitance. This feedback current is 
stepped up in the aerial transformer by a factor of 10-7 ( — 64/6) to 
appear as a feedback current in the coupling coil. Consequently, feed- 
back through the stray capacitance is approximately 70 times more 
important than feedback through the collector-base capacitance of the 
transistor. A stray capacitance of lpF in the tuning capacitor alone will 
be equivalent to about 7 times the internal feedback capacitance of 
the OC44, which is about lOpF. 

The effect of the feedback depends on the phasing of the oscillator 
and aerial transformers. If the phase of the feedback current is such 
that the amplitude of oscillation is increased, squegging may occur. 
Alternatively, if the phase is reversed, the amplitude of the wanted 
oscillation may be decreased. The unwanted feedback is a maximum 
when the receiver is tuned to its highest frequency, and the oscillator 
coil then loses control of the frequency of oscillation, which is then 
determined by the overall properties of the feedback path. The feedback 
is increased when the tuning capacitance has a low value, such as is 
common in miniature tuning capacitors, because of the higher trans- 
former ratios. 

The most practical way of reducing the undesired feedback in the 
circuit is to specify a tuning capacitor in which there is a screen between 
the oscillator and aerial sections, as in this design. Other forms of stray 
capacitance between the oscillator and aerial coils, for example, the 

Page 195 


capacitance of the wiring of the wave-band selector switch, should be 
kept as low as possible. 

Input Resistance of Mixer 

The r.f. input resistance of the mixer transistor does not depend very 
much on whether the local oscillator is functioning. The change in 
input resistance does not exceed 10%, provided the steady emitter 
current is adjusted to be the same for both oscillating and non-oscillat- 
ing conditions. It follows that the input resistance of an average OC44 
may be calculated from values derived from its equivalent circuit. 

The load impedance in the collector is low at radio frequencies. Thus 
it can be shown that the input resistance of a grounded-emitter tran- 
sistor with its output short-circuited to a.c. is given approximately by 

R _ _ , _ iVe+rw 

K ln - r ^'+ rb 'e rb , e+rbb/+t0 2 (cb , e+Cb , c) 2 (rb , e) 2 rbb , ' 

The hybrid iz equivalent circuit gives 
r b b- = 1 10Q, 
gb'e = 390fjimhos, 
Cb'e = 410pF, 
and Cb'c = 10-5pF, 

for an OC44 operating at an emitter current of 1mA and a collector- 
emitter voltage of — 6V. 

The values of gb'e and Cb'e are directly proportional to the emitter 
current. Thus for operation at 0-32mA, which is the current when the 
receiver is tuned to lMc/s, gb'e is 125(jimhos and Cb'e is 131pF. Hence 
rb'e , which is the reciprocal of gb'e , is 8kO. Substituting these values 
in the above equation gives an input impedance of 5k£2 at lMc/s. 

At a frequency of 200kc/s, the4nput resistance of the mixer stage is 
approximately WkQ. for an average transistor. This input resistance 
is controlled not only by the characteristics of the transistor as given 
by the above equation, but also by the bypass capacitance in the 
emitter. This capacitance has a greater effect at lower frequencies. 

Output Resistance of Mixer 

The output resistance at a frequency of 470kc/s varies widely and 
depends on the frequency to which the receiver is tuned. This variation 
results from feedback within the mixer. Without feedback, the output 
resistance would be about 35kQ. 

In practice, the output resistance reaches a minimum several times 
lower than 35kH in the long- wave band, and a maximum several times 
higher than 35kQ in the medium-wave band. The mixer is therefore 

Page 196 


coupled to the first i.f. transistor by an i.f. transformer similar to that 
used for coupling together the two i.f. stages. The output resistance 
with this arrangement is 28kQ. 

This method of coupling makes it possible to standardise the design 
of the first and second i.f. transformers. It also limits the load resistance 
in the mixer output to a suitably low value, so that feedback in the 
mixer is not serious. 

Conversion Gain 

At a signal frequency of 1 Mc/s, the ratio between the r.f. power at the 
input of the mixer transistor and the r.f. power at the input of the i.f. 
transistor is approximately 27dB. 


The basic design of the i.f. amplifier has been discussed in Chapter 18. 
Two OC45 are used in unilateralised greunded-emitter stages. The 
intrinsic power gain of an OC45 at 470kc/s is 38dB and the insertion 
loss of the second i.f. transformer 7dB. The gain of the first stage, 
without automatic gain control, is therefore 31dB. 

The unloaded Q (that is, Q ) of the second i.f. transformer is 100. 
The working Q (or Q w ) is 56, which gives a bandwidth of 8-4kc/s. 
These values of Q and Q w , giving an insertion loss of 7dB, ensure 
that a stability factor of four is maintained in a two-stage amplifier 
containing three similar coils. Thus the loop gain of either i.f. stage 
does not exceed a quarter of the gain needed for oscillation, even when 
the feedback capacitances of the transistor are at the extremes of the 
tolerance range. 

The third i.f. transformer is designed to have an unloaded Q of 160. 
The load resistance on the secondary is arranged to maintain the 
stability factor of four, and also to give a bandwidth of 9kc/s and a 
stage gain of 34dB. 

The identical first and second i.f. transformers are designed for the 
28ki2 output resistance in the preceding stage. However, the output 
resistance of the mixer stage may rise to many times this value, so that 
the stability factor of the first i.f. stage may be reduced undesirably 
from four to three. Although this reduction should not cause oscilla- 
tion, even with the tolerances at their most unfavourable extremes, the 
bandwidth may be somewhat narrower. The required stability factor 
of four can be restored very simply, and at the expense of only about 
2-3dB of gain, by connecting a 750O resistor across the secondary 
of the first i.f. transformer. 

Page 197 







1 1 1 

1 1 1 

I I 1 

1 1 1 







Page 198 



The detector is an OA70 germanium diode in a conventional circuit. 
The steady voltage developed across the diode load resistance (which 
is also the volume control) is fed back to the first i.f. transistor to give 
automatic gain control. 

The small forward bias applied to the diode by the a.g.c. circuit, 
serves to keep the input resistance of the detector constant at all signal 
levels. This small bias gives constant loading on the i.f. amplifier, thus 
preserving the stability factor, and also improves the detection effici- 
ency on weak signals. 


Fig. 5 shows the performance of the simple system of a.g.c. incorpor- 
ated in the circuit of Fig. 1. The range of control is 33dB, and the 









" in 


















60 70 80 90 

Signal input (dB) 
(relative to I uV across aerial) 

Fig. 5 — Performance of a.g.c. arrangement with and without damping diode 

corresponding variation in signal level at the detector is 6dB. (If the 
direct voltage across the diode load resistance is measured instead of 
the audio output, then a better figure for a.g.c. will be obtained.) 

Fig. 4 shows a modification to the a.g.c. system used in Fig. 1 which 
increases the range of control to 52dB. The original feedback path is 
retained, but another diode (an OA79) is included to damp the first 

Page 199 


|£ I I 1 III IN 

Page 200 


i.f. transformer. This transformer is now constructed with a separate 
input winding, so that the tuned winding can be held at a steady 
potential of — 0-22V by connecting it to a point in the emitter circuit 
of the second OC45. 

When no signal is applied, the emitter of the first i.f. transistor has 
a nominal potential of 0-68 V. Thus the reverse bias across the damping 
diode is 0-46V. The diode therefore has a high impedance and its 
effect on the gain of the receiver is negligible. 

If an input signal of increasing amplitude is applied, the a.g.c. voltage 
on the base of the transistor causes the emitter voltage of the first i.f. 
transistor to fall until it becomes less than 0-22V. The damping diode 
is then biased in its forward direction, and its impedance decreases 
rapidly. The diode heavily loads the first i.f. transformer, thus widening 
the bandwidth and allowing a much larger input signal to be handled. 

Fig. 5 also shows the variation in audio output with r.f. input voltage 
for the circuit using the modified a.g.c. system. The fall in output at 
very high signal levels results from clipping of the modulation envelope 
in the mixer. 

For the modified form of a.g.c, the emitter bypass capacitance of 
the first i.f. transistor is increased to 25 [i¥. This capacitance prevents 
a.f. voltages, arising from unwanted detection, from being developed 
at the emitter. Such voltages would cause distortion of the modulation 
envelope at some signal levels. 


The audio amplifier consists of a driver stage and a transformerless or 
'single-ended' output stage. The driver stage incorporates an OC71 
and the output stage a matched pair of OC72 (the 2-OC72). An inter- 
stage coupling (driver) transformer is required, but the loudspeaker 
forms the direct load for the output stage. 

The audio amplifier is designed to have an input impedance of 9k£l 
at the volume control, so that a good a.c./d.c. load ratio is maintained 
for the diode detector. With nominal transistors, a drive voltage of 
200mV is required for an output of 200mW. Negative feedback of 5dB 
is taken from the loudspeaker terminal to the emitter circuit of the 
OC71 driver. 

Higher Output Powers 

Audio amplifiers consisting of the LFH3 packet (an OC81D driver 
transistor and a matched pair of OC81 output transistors) may be 
incorporated if desired. These circuits were described at the end of 
Chapter 14 (Figs. 10 to 13). According to the form of circuit adopted, 

Page 201 


output powers of up to 1W are obtainable. For all four circuits, the 
resistance in series with the connection from the volume control should 
be increased from 5-6kQ. to about 8-2kQ. 

Modern commercial receivers normally provide a much higher 
output power than is obtainable from the 2-OC72. 


The circuit of a semi-miniaturised version of the receiver is shown in 
Fig. 6. Basically this modified version and the original are very similar. 
However, the output power of the miniature version is reduced to 
lOOmW (as compared with the performance given on p. 203) and 
smaller components are used. The current consumption is also reduced 
and is 7mA with zero signal and 13mA with average programme. 
The miniature receiver has been made with overall dimensions of 
6|x4£xl| in. ( ~ 16-5xllx4cm) and further reductions in these 
are possible. 

The frequency range is modified to 550 to 1600kc/s in the medium- 
wave band and 150 to 250kc/s in the long- wave band. The lowest 
possible stray capacitance is essential in the wave-band selector switch 
in the oscillator section. To avoid switch capacitance in the aerial 
circuit, the selector switch has been transferred to the low end of the 
medium-wave coil, so that both coils are in series for long-wave recep- 

In small personal receivers, the aerial is small and is almost always 
damped by adjacent metal parts. Thus two i.f. stages are still required 
if the receiver is to have sufficient sensitivity for use throughout Great 
Britain, for example. 

Similarly, because the loudspeaker in a miniature receiver is small 
and relatively inefficient, no great reduction in output power is possible. 
The driver and class B stages are therefore still required, especially 
since battery economy is of major importance when battery size is 
reduced to a minimum. Nevertheless, it is advisable to reduce the out- 
put power to lOOmW in order to limit vibration in the receiver case, 
and also to limit the battery consumption. 

When low initial cost is more important than battery fife, and a 
limited output is acceptable, a class A output stage could be used. 
Thus the number of transistors may be reduced to five. Another 
suggestion for reducing the number of transistors is to use a reflex 
arrangement in which the last i.f. transistor also acts as the first audio 


For enlarged receivers intended for permanent use in the home, the 

Page 202 


existing design can easily be adapted, with very few modifications. 
A large Ferroxcube rod can be incorporated to give maximum sensi- 
tivity. Also a highly efficient loudspeaker can be used. With such a 
speaker, the volume obtained from the existing output of 200mW 
should be adequate for most listening conditions, while retaining a low 
battery consumption. The life of a moderately large battery could be 
expected to be about a year. 

Higher output powers may be provided as just described where more 
frequent battery replacement is not a disadvantage. 


Output power 200mW 

Frequency range 

medium-wave 540 to 1640kc/s 

long-wave 155 to 280kc/s 

Frequency response . . . . 6dB below response level at lkc/s at 70c/s and 3kc/s 

Sensitivity 200[lV signal (30% modulation at 400c/s) across aerial tuned circuit for 

audio output of 50mW 

adjacent-channel rejection ratio at 9kc/s off-tune 29dB 

second-channel rejection ratio 50dB 

A.G.C. range 
change in input signal for 6dB change in audio output 

(a) without damping diode 33dB 

(b) with damping diode. . 52dB 

Battery consumption 

zero signal . . 9mA 

average programme . . 20mA 


Printed circuit: Weymouth Radio Manufacturing Co. Ltd. PCA1 


Mullard Ferroxcube rod FX1268 

Primary 64 turns of 19/0-0028 bunched conductors wound in single layer 

(Qo = 200 at lMc/s) 
Secondary 6 turns of 19/0-0028 bunched conductors wound at low end of 

primary coil 


Primary 175 turns of 34 s.w.g. rayon-covered enamelled wire, wave wound 

(Qo = 80 at 200kc/s) 

Secondary 41 turns (continuation of primary coil) 

Complete aerial: Weymouth Radio Manufacturing Co. Ltd. RA2W 

Page 203 


Tuning Capacitor 

Aerial section 175pF max 

Oscillator section 123pF max 

Plessey type W with inter-section screen 

Oscillator Coil (screened) 

Inductance 173fxH 

Windings . . . . . . . . Main = 72 turns 

• Collector = 1 1 turns 
Emitter = 2 turns 

Qo^ .. 1 30 at fosc = lMc/s with tuning capacitance of 132pF 

Weymouth Radio Manufacturing Co. Ltd. P50/1AC 

I.F. Transformers 

The leakage inductance in the coils in the i.f. transformers should be kept as low 
as possible. Enclosed pot cores are desirable. 


Tuning capacitance (referred to collector) 3000pF 

Ratio of collector winding to secondary winding . . 6:1 

Qo 100 

Q w (in circuit) nominal 56 

Weymouth Radio Manufacturing Co. Ltd. P50/2CC 

♦If a damping diode is used in the a.g.c. system, a separate collector winding on the first i.f. transformer 
is necessary. Otherwise, this transformer is identical with the second i.f. transformer. Weymouth 
Radio Manufacturing Co. Ltd. (type no. not available) 


Tuning capacitance (referred to collector) . . 4000pF 

Ratio of collector winding to secondary winding .. 1-85:1 

Qo 160 

Q w (in circuit) 52 

Weymouth Radio Manufacturing Co. Ltd.P50/3CC 

Driver Transformer 

Turns ratio 3-6:(l + l) 

Primary inductance . . 5H at 2mA d.c. 

Primary resistance less than 200Q 

Secondary resistance .. . . less than 60£1 per winding 

Fortiphone A443 


Speech-coil impedance . . . . . . 350 

Elac 5in. 5D/211 ; 3in. P/319; 4 x 7in. 47D/108. 
TSL (type no. not available) 

Page 204 



Voltage (4-5+4-5)V 

Ever Ready Batrymax D 


Aerial, oscillator coil, and aerial and i.f. transformers: 

Weymouth Radio Manufacturing Co. Ltd. (type nos. not available) 


Mullard Ferroxcube rod FX1057 


Primary . . 88 turns of 19/0-0028 bunched conductors wound in single layer 

Secondary 1 2 turns of 1 9/0 0028 bunched conductors wound at low end of primary, 

nearest centre of rod 


Primary . . 240 turns of 3 /0 0024 bunched conductors wave-wound in three sections 
Secondary 45 turns of 3/0 0024 bunched conductors wound as fourth-pie section, 

nearest centre of rod 

The total length of the four l.w. windings (including three spaces of ift-in. each) 

is one inch (~ 2 -5cm) 

Tuning Capacitor 

Aerial section 115pF swing 

Oscillator section 11 5pF swing 

Wingrove and Rogers C78-22 or C78-02 (with slow-motion drive). Both types 

to include an inter-section screen. 

Oscillator Coil 

Inductance 313(xH 

Windings Main = 100 turns 

Collector = 13 turns 

Emitter = 2 turns 

Qo 105 at lMc/s with tuning capacitance of 70pF 

I.F. Transformers 

As given for original version 

Driver Transformer 

Turns ratio . . . . . . . . 7 ; (| _|_i) 

Primary inductance 5H at 1 -5mA d.c. 

Primary resistance less than 750Q 

Secondary resistance less than 100ft per winding 

Fortiphone L442 

Page 205 



Speech-coil impedance 75fi 

TSL 2in. CMS50 


Voltage 9V 

Ever Ready Batrymax PP4 

NOTE: The inclusion of a type number should not be taken to imply either that the list is exhaustive 
or that specific components are generally available. 

Page 206 



The transmitter described in this chapter is capable of delivering 
an output of 4W into a resistive load, and has been designed to operate 
at the international marine distress frequency of 500kc/s. This frequency 
is, of course, strictly reserved for the use of those in distress, but the 
design procedure is fully described and the circuit can easily be modi- 
fied for neighbouring frequencies. 

Modulation arrangements are described for manual keying and radio 

Apart from the interest of the transmitter as a whole, it illustrates 
many of the applications described in other chapters, plus some new 





Push-pull power 



A.F. Wor Morse 

Fig. 1 — Block schematic of transmitter 

ones. The design includes a crystal-controlled oscillator, a silicon 
transistor used other than as a d.c. amplifier, and Ferroxcube pot cores. 


A block schematic of the 500kc/s transmitter is shown in Fig. 1. The 
transmitter comprises an oscillator, an r.f. amplifier, a driver and a 
power amplifier. 

The oscillator is crystal controlled and uses an OC45. The resonant 
circuit in the collector lead is tuned to the series-resonant frequency 
of the crystal. The crystal is connected in series with the feedback 
winding to control the frequency of the oscillation. 

The r.f. stage is an OC201 silicon transistor. The output from this 
stage is applied to a driver amplifier and then to a push-pull power 

Page 207 


amplifier which feeds an aerial. Both the driver and the power amplifier 
use OC24 high-frequency power transistors. 

The r.f., driver and power-amplifier stages are operated under class B 
conditions. Class C operation could be adopted, but the modulation 
would have to be applied to the output stage and, since a higher modu- 
lator power would be required, there would be little or no improvement 
in the efficiency of the transmitter as a whole. Class C operation of 
transistors is in any case not particularly attractive ; hole-storage effects 
reduce the efficiency considerably below the theoretical value, and the 

R2 < C2 
4-7< O-Olj 

Fig. 2 — 4W transmitter for 500kc/s 

variation in current gain with frequency and current give a further 
departure from the ideal. These effects would probably necessitate an 
empirical design. 

The main circuit of the transmitter is given in Fig. 2. 

Collector Dissipation 

The amplifier has been designed and tested for thermal stability 
at a maximum operating ambient temperature of 60°C (140°F) and at 
a maximum operating junction temperature of 75°C. 

Suitable heat sinks for the two OC24 transistors in the push-pull 
amplifier can be made from is-m. ( — 16 s.w.g.) aluminium sheet. The 
surface should be blackened to assist cooling by radiation. Each tran- 
sistor should be mounted centrally on its own heat sink of area 5 x 6 in. 
( ~ 12-5 X 15cm). The thermal resistance of the heat sinks should be 
about 4-4°C/W. The transistors should be mounted directly on the 
heat sinks, and the heat sinks insulated from the chassis. 

Page 208 


In practice it may be more convenient to use the chassis as a common 
heat sink, the transistors being insulated from the chassis in the usual 
way. The mounting-base temperature of the transistors, for a maximum 
operating ambient temperature of 60°C, should then not exceed 72°C. 

Collector Load 

In a class B push-pull amplifier, the mean ax. power P ou t is given by 

p (Vr.m.B.) 2 

rout = 


where R ce is the load resistance applied to each transistor in turn, 
and Vr.m.s. is the r.m.s. value of the output voltage. 

The design of the class B push-pull amplifier follows the same basic 
technique as that for an audio amplifier except that, since the collector 
circuit is tuned, no bias is required to eliminate crossover distortion. 
If a 12V battery is used to supply the collector circuit, and allowing 
0-5 V for bottoming, the output voltage at the collector is 

i^L 5 = 8.1V„„,. 

The collector-emitter load resistance per transistor is 

Rce = (V r . m ...)2/Pout = ^ = 16-4Q. 

The use of a shared emitter resistor reduces the effect of variations in 
the base-emitter voltage V be , and reduces the lengthening, caused by 
hole storage, of the collector-current pulse. The reduction of the hole- 
storage current gives a marked improvement in the efficiency of the 
amplifier. The resistance in the emitter lead must not be too high, 
because it reduces the gain of the stage. With an emitter resistor of 
IQ, the collector load becomes R c = 16-4—1 = 15-4D. The collector- 
collector load R c -c is then 4R C = 61-6Q. 

Output-transformer Design 

To avoid loss in the tuned circuit, the working Q of the circuit should 
be as low as possible, though too low a Q results in greater harmonic 
content. A Q of 10 or 20 is recommended. In the design of the output 
transformer, using a Ferroxcube pot-core assembly LA6, a working Q 
(Q w ) of 15 is assumed with an unloaded Q (Q ) of 230. Because 
Qo > Qw , the dynamic impedance of the circuit at resonance is 
effectively equal to R c , the collector load. Thus 

Rc = &>LQ W 

and L = R c /ojQ w . 

Page 209 


For R c = 15-4Q at 500kc/s, 


L = o Ag iA fi k = 0-327{xH. 

2ttx0-5x10 6 x15 

The number of turns required for 0-327(xH using a Ferroxcube pot-core 
assembly LA6 is 

m = 92 VL [mH] = 92 V(0-327 x lO" 3 ) = 1 -66 ~ 2 turns. 

Q w cannot be increased because the winding would be less than 2 turns. 

If ni = 2 turns, the new value for L is 

L=(Q 2 [mH] = g) 2 = 0-472^H. 

The capacitance to tune the collector to resonance at 500kc/s is 

C - 1/M2L = (2 tc x0-5x10«)'x0472x10-« = ^^ 

In practice, it is impossible to obtain a high value of Q with only 
two turns ; therefore a tertiary winding is used for tuning. If the number 
of turns across the transistor output is ni , and Ct is the tuning capacit- 
ance for the tertiary winding, then the number of turns n3 required 
for the tuned circuit is 

n 3 = niV(C/C T ) . 
The highest practicable tuning capacitance is used to avoid high voltages 
across the circuit. If the tuning capacitance Ct is 820pF, 

113 = 2 /( 820x10-12 ^ 33tUrnS - 


To feed an aerial of impedance R ae , the turns ratio m/n2 between 
the collector and aerial is chosen to obtain a reflected load of 61-6Q 
for R c -c or 61-6Q/4 for one transistor. Alternatively a turns ratio of 
1:1 or 1:2, in conjunction with an impedance-matching network, may 
be used. This can take the form of either an L or a tt section. 

The Tt network shown in Fig. 3 is for matching the output impedance 
R of the transmitter to the aerial impedance R ae • The values of Xa, 
Xb, and Xc can be found from 

X A = X C = -V(RxRae) 

and X B = V(RxRae). 

The values of L and C are given by 

T _ ^ B 

L_ 27rf 

and C = - • 


Page 210 


The advantage of such a network is that the reactance X c can include 
the capacitive reactance of the aerial. The efficiency of the matching 
network is of the order of 90 %. 

A resistive load and a turns ratio of 1 :2 may be used for checking 
the transmitter. If ni = 2 turns and n 2 = 4 turns, the preferred value 
of 68Q±10% may be taken for the resistive load of 61-5Q. 


The driver stage Tr3 is also operated in class B. This stage has to provide 
an output power of about 500mW, and operates at less than about 
25 % of the dissipation of each of the output transistors. The heat sink 

R — X A? 

yXg - — Roe 

Fig. 3 — 7t network for matching aerial 

therefore needs to be only about 25% of the area of theirs, that is, 
about 2|x3 in. ( ~ 6-5x7 -5cm). If the transistor is mounted on the 
chassis, with suitable insulation, the mounting-base temperature 
should not exceed 74°C, for a maximum operating ambient tem- 
perature of 60°C. 

The input impedance Z in of the output stage, at higher emitter 
currents, is very nearly rut,- (70Q). Because the driver amplifier is 
working at a higher level of impedance than the output stage, Q w 
may be increased. If Q w = 20, the inductance required for the base 
winding is 

' V 2tc X0-5 X 10 6 X20 m ' 

and the number of turns, using Ferroxcube pot-core assembly LA3, is 

n 2 = 85 VL [mH] = 85 y/(l -12 x 10~ 3 ) ~ 3 turns. 
The input power required to drive the push-pull amplifier using low- 
gain transistors was measured as 500mW. The input power is given by 

(Vin(r.m.s.)) 2 


so that 

V ln = A/(PinZ in ) = V(0-5 X70) = 5-9V r .m. s . . 

If ni is the number of turns required at the collector of the driver 

Page 211 


ni = — -n 2 - 


Allowing 0-5V for bottoming, V c is equal to 8-1V r.m.s., and 

8-1x3 . + 
m = — j-£— ^ 4 turns. 

A slightly higher driving voltage is obtained at the input to the push-pull 
amplifier with the approximated winding of 4 turns. If required, 
the input power may be adjusted by means of the variable emitter 
resistor RV4 at the input to the driver amplifier. A tertiary winding is 
again used for tuning the collector current. 

If 4 turns are used for ni , the new value of inductance is 

L = © 2 [mH1 = © 2 = 2 ' 22 * H - 

The tuning capacitance is: 

C = 1 /»* L - (2,xO-5xlOVx2-22xlO-e = 00457 « R 

If Ct = 820pF, the number of turns required for the tuned winding 
is given by 


„„,„ , A //0-0457xl0- 6 \ , A 
m V(C/C T ) = 4^ ( 820xl0 _ 12 ) - 30 turns. 

An emitter resistor of 2-2Q, is required to reduce hole-storage effects 
and increase the efficiency of the amplifier. 


The design procedure is similar to that in the previous section. The r.f. 
amplifier is also designed for class B operation. A 6V supply is used 
because of the voltage limitations of the oscillator transistor and 
because it is convenient for the modulator; it is also used because of 
the keying system adopted. 

The input power required by the driver amplifier was 35mW at 
2-3V r.m.s., measured with a low-gain transistor. The average input 
impedance is 

Zin ~ — P~~ - 35^10^ - 15ia 

As the stage is working at a higher level of impedance than the driver 
stage, the working Q can be further increased. A Q w of 50 was chosen 
for good harmonic rejection. 

The inductance required for the base winding of the r.f. amplifier is 

Page 212 


L = ZW..Q. = ^o-sxUxSO = °' 96 ^ H - 

For this stage a Ferroxcube adjustable pot-core assembly LA35 is 
used. The number of turns required for the base winding is given by 

n 2 = 135VL [mH] = 135V(0-96xlO- 3 ) ~ 4 turns. 
The available voltage swing, allowing for the voltage across the tran- 
sistor and emitter resistor, is approximately 3 -25V r.m.s. 

The number of turns ni required at the collector is found from 

_. ... 4x3-25 . 
ni = n 2 .Vc/Vin = — ^t~ - 6 turns - 

The inductance of the collector winding is 

and the tuning capacitance is 

C ^ 1/w2L = (frxO-Sxloj'xHWxlfr- = °° 513(iF - 
This high value of capacitance is rather inconvenient, so some extra 
turns are added to the collector winding and the tuning capacitance 
reduced to Ct • With Ct equal to 680pF, the total number of primary 
turns is 

„^,^s r //0-0513xl0- 6 \ M 
n 3 = ni V(C/C T ) = 6 N /( 68Qxl0 _ 12 j * 52 turns. 


A crystal-controlled oscillator with a tuned collector circuit is used. 
Feedback is obtained by a transformer winding which couples the 
parallel-tuned circuit in the collector to the base. The crystal is connected 
in series with the feedback winding, and the collector circuit is tuned to 
the series-resonant frequency of the crystal. The series resistance of 
the crystal should be sufficiently low for the feedback current to start 
oscillation. For the feedback to be of sufficient magnitude and the loop 
gain greater than unity, the turns ratio of the transformer should be 3 :1. 

The oscillator works under class A conditions. Base bias is provided 
by a potential divider R1-R2, in conjunction with an emitter resistor 
R e , which is bypassed to r.f. 

The oscillator is an OC45. The circuit is designed to drive the OC201 
r.f. stage. This amplifier requires an input of 1 -3V r.m.s. for a transistor 
having a minimum a' of 20. 

Page 213 


The oscillator has a nominal quiescent collector current of 2mA. 
The available voltage swing, taking into account the voltage drop 
across the emitter resistor, is approximately equal to 3*4V r.m.s. 

If Ti2 , the base winding for the OC201, is 7 turns, the number of 
turns at the collector is 

ni = n 2 .V c /Vb = ~ 18 turns. 

Therefore 6 turns are required for the feedback winding if a 3:1 turns 
ratio is used. 

With a Ferroxcube adjustable pot-core assembly LA35, the induct- 
ance of the collector winding is 

±y imin -mx 

135/ [mH] - 135 = "■ 8 « JI - 

The capacitance required is 

C = 1/w2L = (2»xO-5xloj'x 17-8x10- = 5680pF - 
As before, extra turns are added to the collector winding, so that a 
smaller tuning capacitance, Ct , of 680pF may be used. 

The total number of collector turns now required is 

,/~/~ x 10 //5680xl0-i2\ ^ 
n 3 = ni V(C/C T ) = !8 > /( 6g0xl0 _ 1 , ) - 52 turns. 


The procedure adopted is the usual one of rough tuning at reduced 
power, followed by final adjustment at full power. 

A resistor of 27Q is connected in the emitter lead of the r.f. amplifier, 
Tr2, in order to reduce the available input to the driver amplifier. 
(RV4 may be set to the middle of its range for this purpose.) The 12V 
supply is then connected to the transmitter, and the oscillator tuned 
circuit is adjusted to the series-resonant frequency of the crystal, 
by adjusting the inductance of the pot core. The r.f. amplifier is tuned 
next, again by adjusting the pot core. The driver and the power- 
amplifier stages are then tuned for maximum output voltage across 
the 68Q load, by adjusting the tuning capacitance. (Use a fixed capacit- 
ance of just below the calculated value and shunt this with a variable 
trimmer.) An output of about 2W should be obtained. 

The emitter resistor of Tr2 should then be short-circuited. The 
complete transmitter should be readjusted, starting with the oscillator, 
followed by the r.f. amplifier, the driver stage, and finally the power 

Page 214 


amplifier. The tuning of the power amplifier should be carried out 
carefully, as excessive collector current flows when the stage is off 
tune. The full output power of 4W should now be obtained. 

RV4 is subsequently used to set the output power to the desired level. 


An experimental transmitter has been built using the circuit shown 
in Fig. 2. The performance has been tested over a range of ambient 
temperature from to 60°C. The variation of output power, measured 
across a resistive load of 68O, is shown in Fig. 4. From the graph, 
it can be seen that the output is higher at lower temperatures, and falls 





6 4 " 




(a? -^^ 


Fig. 4 — Variation in output from transmitter with temperature 

gradually at temperatures above 25 °C. Quite reasonable output 
power is obtainable even at 60°C. The reduction of the output 
power arises from the fall in a cut-off frequency and the increase in 
hole-storage current with temperature. The latter can be seen from an 
expression for the hole-storage time-constant, 

1 — oiNai 
(where n denotes normal and i inverse). 

Because w falls and a rises with temperature, there is a rise in the 
time-constant t s with temperature, which accounts for the increased 
hole-storage current. 

Curve (a) in Fig. 4 is for low-limit transistors. With typical transistors, 
the output and driver stages may be overdriven at low temperatures, 
in order to obtain increased output at higher temperatures. Overdriving 
is achieved by selecting transistors with higher gain for the r.f. amplifier 

Page 215 


and driver amplifier, or by adjusting the turns ratio. The variation in 
output power for high-gain transistors is shpwn in curve (b) in Fig. 4. 
The variation in frequency has been found to be less than one part in 
10 6 per °C over the entire range of temperature. 

Some field tests have been carried out at 520kc/s. The only modifica- 
tions to the design meant using a 520kc/s crystal for the oscillator and 
retuning the transmitter to the new frequency. A transmitting aerial 
30 ft. high ( ~ 9m) and a receiving aerial of about 30 ft. of wire were 
used. Strong signals were received within 12 miles ( ~ 19km) over land. 
This is not the maximum range over land, and a range of at least 
50 miles ( ~ 80km) may be expected at sea. 


Low-level modulation is used for radio telephony. The r.f. amplifier 
Tr2 is biased into class C operation, using a parallel CR combination, 

Fig. 5 — Modulator for radio telephony 

by the normal rectification of the emitter diode. The bias is adjusted 
to the point where the output of the transmitter falls to half its normal 
value, and an a.f. signal is applied across the bias resistor (Fig. 5). 

Manual Keying 

A circuit for on-off keying is given in Fig. 6. A morse key is connected 
in series with the base winding, and an r.f. choke and a bypass capacitor 
are added to prevent r.f. voltages from floating across the key. 

The r.f. amplifier is thus d.c. keyed in its base circuit. This keying 
point is possible because, although with the base open circuit the leak- 
age current assumes its full value of I'co , the value of I'co is very low 
for a silicon transistor, as used for this stage. 

Page 216 


Automatic Keying 

The morse key in Fig. 6 may be replaced by an automatic key which 
will give continuous unattended operation. The transistor multivibrator 
provides a simple switching element around which to design such a 

Automatically-keyed transmitters are of interest to the whaling 
industry. A transmitter can be left aboard the harpooned whale, which 
is collected later when the kill is complete. The circuit is modified to 




Fig. 6 — Modulator for manual keying 

operate at a fixed frequency other than 500kc/s, and is provided with 
an automatic-keying device which generates the simplest convenient 
signal. Such transmitters may be operated on continuous wave or 
modulated continuous wave. 

Another application of automatic keying is to the generation of the 
S.O.S. distress signal. The signal to be generated consists of the group 

• : • • • •, transmitted as a single signal with the dots sufficiently 

distinguished from the dashes. A minimum of four multivibrators and 
a gate would be needed to generate the signal. When intended for 
distress-signal transmissions, the transmitter may be designed for 
500kc/s, but should be modified for modulated-continuous-wave 
operation, in order to comply with the international requirements. 

Continuous Wave (C.W.) or Class A1 Emission 

With the arrangement shown in Fig. 6, the transmission consists of the 
r.f. carrier interrupted by the on-off key. As there is no audio modula- 
tion, the result is continuous wave (c.w.) or class Al emission. 

Modulated Continuous Wave (M.C.W.) or Class A2 Emission 

The modification for modulated-continuous-wave operation entails 

Page 217 


providing an auxiliary audio oscillator to modulate the r.f. carrier. 
It is permissible to key either the audio modulation alone, or the r.f. 
carrier and the audio modulation together. One possible system for the 
latter method (Figs. 2 and 6) is to apply the keying signal in series with 
the audio modulation to the base winding of Tr2. 


The transmitter is particularly suitable for lifeboats and life-rafts, 
but can be adapted for life-jackets, air-sea rescue equipment, helicopters, 
and radio beacons for coastal stations. The transmitter may also be 
of interest to climbing and scientific expeditions. 

A 4 ampere-hour accumulator (the Exide 6MNA17, for example) 
will power the transmitter and an automatic-keying device for some 
50 hours. 

Oscillator Transformer T1 








Winding : 

Ferroxcube adjustable pot-core assembly LA35 

London Electric Wire Co. and Smiths Ltd., 9/0024 bunched conductors 
Primary Secondary Tuned Feedback 





n 4 

Modulator Transformer T2 

Ferroxcube adjustable pot-core assembly LA35 

London Electric Wire Co. and Smiths Ltd., 9/0024 bunched conductors 
Primary Secondary Tuned 






Driver Transformer T3 

Ferroxcube pot-core assembly LA3 

Connollys Ltd., 3 strands of 19/0028 bunched conductors (m, n2> and 

19/0028 bunched conductors (m) 

Primary Secondary Tuned 



n 2 



Output Transformer T4 

Ferroxcube pot-core assembly LA6 

21 s.w.g. enamelled copper (m, n 2 ) and Connollys Ltd., 19/0028 bunched 
conductors (n3) 

Primary Secondary Tuned 






Page 218 



A sinusoidal oscillator may be regarded as a tuned amplifier, with part 
of the output signal fed back to the input such that the output signal 
is maintained. Oscillation will occur if both the following conditions 
are satisfied: (a) the phase of the feedback signal is such that, after 
passing through the amplifier, the amplified feedback is exactly in phase 
with the output from which it is considered to be derived ; and (b) the 
new output is equal to or greater than the original output. These two 
conditions mean that the loop gain of the circuit must be real and equal 
to or greater than one. 

For oscillations to build up from zero, practical oscillators have a 
loop gain greater than one. On switching on the oscillator, the oscilla- 
tions continue to build up until the effective gain is somehow reduced 
so that the loop gain becomes one. 

The reduction in gain generally occurs in one of two ways : either 
the amplifier bottoms, thereby reducing the load resistance; or else 
part of the sinusoidal signal is converted to a bias, which is arranged 
to cut off the amplifier for part of each cycle, and thereby reduce the 
effective transfer conductance. 

Bottomed oscillators have been rather out of favour in the past. The 
minimum gain of a modern transistor such as the OC72 or OC84, 
however, is sufficient for these oscillators to be designed to work with 
all transistors of the type in question, and at fairly low distortion even 
when a maximum-gain transistor is inserted in the circuit. 

The bottomed and unbottomed oscillators described in this chapter 
are designed in a similar manner to small-signal transistor amplifiers. 
The oscillator performance can be predicted from the mean values of 
collector/emitter current, a' and input capacity. In particular, the out- 
put voltage can readily be calculated. 


Fig. 1 shows the bare essentials of the output circuit of an oscillator. 
The transistor collector current, which generally consists of pulses, is 
set to a mean value I m ean by some suitable bias circuit. (For the pur- 
pose of this chapter, the mean emitter current can be taken to be equal 

Page 219 


to the mean collector current. These will be referred to indifferently 
as Imean •) Usually the bias arrangement takes the form of either a 
resistor connected between the base of the transistor and the negative 
side of the supply, or a base potential divider, which defines the voltage 
at the base and, in conjunction with a bypassed emitter resistor, defines 
the mean emitter current. 

The peak output voltage of the oscillator, v ou t(pk) , is related to the 
mean current by the feedback applied, and will have a maximum value 

2 imean I ^L I 

where |Zl| is the magnitude of the load impedance at the frequency of 
oscillation. In practice v ou t(pk) will never be more than 30% less than 

Fig. 1 — Output circuit of oscillator 

this value in any acceptable design. If, as is generally the case, the 
oscillator is working at the resonant frequency of the tuned circuit, the 
load impedance is simply Rl, and the peak output voltage will be 
approximately (within 30 %) equal to 2 X Imean Rl . 

The value of Rl can be determined for the desired output power once 
the maximum permissible value of v ou t(pk) has been decided. From first 

t, _ (Vout(pk)) 2 

If the oscillator is allowed to bottom, and assuming no emitter resistor, 
Vout(pk) will be approximately equal to V C c . 

Page 220 



The above approximation, that 

Vout(pk) = 2I me an Rl , 

becomes more accurate as the collector-base feedback (and hence the 
loop gain) is increased, although the feedback cannot be increased 
indefinitely because of*squegging. Again the d.c. bias circuit does not 

Fig. 2 — Simplest bias circuit 

affect the general principles of operation, and there are several ways of 
applying the bias. Fig. 2 shows the simplest bias circuit with a base 
resistor Rb . 

An unbypassed resistor R f may be included in the emitter to introduce 
a.c. feedback, but it should be remembered that in some circuits, 
R f = 0. 

The magnitude of the collector-base feedback can be assessed from 
the loop gain A , which is calculated on a small-signal basis using mean 
values of the transistor parameters. A is defined as the product of the 
forward and reverse voltage-transfer ratios. 

The forward voltage transfer is g m RL , where g m is the transfer (or 
mutual) conductance. The reverse voltage transfer is simply the trans- 
former ratio 1/n; the loading effect of the transistor input on the 
secondary is small and can usually be neglected. 

Thus the loop gain is 


„ 1 



• d) 

If the internal base resistance is ignored, then substituting 

a 1 

gm = — ^ — 

r e r e 
Page 221 




A = -^— x-> •••da) 

(r e +Rf) n 

where R f is the external unbypassed resistance, if any, in the emitter 

The peak output voltage, which has been ^iven as 2I mea n Rl , is 
strictly speaking equal to RLXipk(fund) , where ipk(fund) is the peak 







V bic 


/(Rf+r e )T 

---I" - 




1 1 
1 1 








Loop gain Aq 

Fig. 3 — Design curves for Rr greater than zero 

collector current at the fundamental frequency of the oscillation con- 
tained in the collector-current pulses. The value of i P k(£ U nd) can be 
found from Fig. 3 for circuits which include a finite feedback resistance 
Rf , or from Fig. 4 for circuits in which Rf = 0. 
For instance, if the loop gain A is 3, then from Fig. 4 (for R f = 0), 


= 0-9, 

instead of 1-0, as required by the approximation made above. For a 
mean current of 1 -5mA, the peak collector current at the fundamental 

Page 222 


frequency is 

2x1-5x0-9 = 2-7mA 

instead of approximately 3mA, and if the load resistance Rl is 4kQ, 
the peak output voltage will be 10-8V, instead of 12V. 

Only the fundamental component of the peak collector current need 





/ I mean 

. _ 





Loop gain A 

Fig. A — Design curves for Rf equal to zero. Bias voltage in units of kT/q, 
where kT/q ~ 25mV at Tamb = 298°K (25°C). 

be considered, because the second and higher harmonics are bypassed 
by the capacitor in the tuned circuit. 


The difficulty presented by the internal base resistance r b b- is that, 
unless this is only a small fraction of the input impedance of the 
transistor, an appreciable amount of the feedback will be lost. Since 
rbb' , and hence the loss, varies from transistor to transistor, so will 
the output from the oscillator. 

Low Frequencies 

At low frequencies, the loss arises because a' r e is not infinite, and the 

Page 223 


input voltage is tapped down in the ratio 

vm __ a'o(r e +Rf) ...(2) 

e s ~ r b b'+Rf+a'o(r e +Rf) ' 
as is represented by the equivalent circuit of Fig. 5. This attenuation 
factor can be reduced by operating with small mean currents and there- 
fore large r e , or else it can be controlled to some extent by means of the 
feedback resistance Rf . The value of Rf is typically made such that 

a' (r e +Rf) = 3(rbb'+Rf). 
Rf raises the input impedance of the stage more than it increases the 
resistance effectively in series with the base. 

The attenuation factor can be taken into account when calculating 
the reverse voltage transfer and the loop gain A . The forward voltage 

Fig. 5 — Equivalent circuit of oscillator for low frequencies 

transfer also will be affected by Rf , since as Rf is increased, g m falls. 

High Frequencies 

At high frequencies the problem is a little more complicated, but may 
be treated in a similar manner. The shunting of the input by the emitter 
capacitance of the transistor, c e , is shown in Fig. 6. The input impedance 
of a' (r e +Rf) in parallel with the reactance of c e must now be made 
three or more times greater than (r b b'+Rf). This condition may be 
satisfied by a suitable choice of Rf . However, since 

r e will have a sufficiently high value at low mean currents for R f to be 
zero. The exact calculation of the loop gain is more difficult at high 
frequencies because of the phase shift introduced by the input capacity. 
The situation is simplified if the operating frequency is sufficiently 
high for only the reactance of c e to be considered. The input capacity 
c e causes a phase shift between the input voltage vm and the voltage 
returned by the transformer, e s . Because an oscillator always works 
with a real loop gain (zero phase shift), there will be an equal and 
opposite phase shift in the collector tuned circuit, and the load, equal 

Page 224 


to |Zl|, will be somewhat less than Rl. It happens that the same 
proportion of voltage is lost at the output as at the input. The reduced 
value of the loop gain A can therefore be calculated using the 
attenuation factor 

(X ce ) 2 

(rw+Rf) 2 +(X ce ) 2 

From Fig. 3 or Fig. 4, the value of the peak collector current at the 
fundamental frequency can be found which corresponds to the reduced 
A . 

The poorest transistor is the one having the largest input capacity, 
that is, the lowest on , since 



coi(r e +Rf) 
The output voltage with this transistor is less than for a nominal 

r J VWv WW- 

rfcbf Rf 

(£\Vs a'o(r e +*f>! c e X 

C e*U)|Cr e +R f ) 9 ™= (r e+ R f ) 

Fig. 6 — Equivalent circuit of oscillator for'high frequencies 

transistor, and will have a value given by 

reduced v ut<pk) = reduced RlX reduced ipk(fund) , 
the reduced Rl being found from 

Rl • X C e 

VUrw+Rf^+CXce) 2 } " 

Maximum Value of R t 

It can be seen that the value of Rf is decided by the internal base resist- 
ance rbb' , by a', and by the frequency of operation, but is independent 
of the required output power. 

At the higher power levels, the load resistance becomes comparable 
with Rf , so that the maximum output power is limited, unless the supply 
voltage is increased. This limit can easily be illustrated by referring to 
the graphs. For instance, at a loop gain of 2, the peak collector/emitter 
current is about 3 times the mean (Fig. 3), and so when Rf is equal to 
one-third of the load, the same voltage, though of different waveform, 
is developed across both Rf and Rl . The circuit is then not capable of 
more than 50 % efficiency. 

Page 225 


Base-resistor Bias 

With the simple bias arrangement shown in Fig. 7, the mean current is 
approximately equal to the mean a' times the mean base current, that 

IS, Imean — a' X lb (mean) . Since Ib(mean)Rb — V cc , 


Rb ^ 


where the leakage current is ignored. From the known value of a', 
the mean current can be set to the desired value by choosing a suitable 

Fig. 7 — Schematic oscillator with base-resistor bias 

value for Rb . This arrangement is not very satisfactory as it stands, 
because a' varies from transistor to transistor and so introduces a 
spread into the mean current for any fixed value of Rb . 

However, the circuit works very well if the value of Rb is made equal 
to that which will just cause the lowest-a' transistor to bottom 
(vout(pk) = V cc ). Higher-a' transistors are then more heavily bottomed, 
so that their mean a' is almost reduced to that for which the circuit 
was designed, and there is little change in the mean current. For the 
same reason, the circuit is not affected by the increase in I co with 
temperature, except at very low currents. 

Circuits can be designed on this basis to give distortions of only 
1 % with the most heavily bottomed transistor. 

The mean base current when the oscillator is working is higher 
than the direct current in the quiescent state, on account of the bias 
voltage developed across the decoupling capacitor 0>. This voltage 
is in a direction such that it tends to increase the mean base current, 
so that for complete accuracy the bias voltage should really be added 
to Vcc when calculating Rb . Also, the mean base-emitter voltage 

Page 226 


Vbe+Imean Rf should be deducted from V cc • The design equation 
therefore becomes 

Rb = 

(Vcc+Vbias— Vbe~ ImeanRf) Xa'mtn 

The bias voltage can be found by means of Fig. 3. 

Bias by Potential Divider and Emitter Resistor 

In oscillators which are to be modulated (for instance, self-oscillating 

Fig. 8 — Schematic oscillator for potential-divider and emitter-resistor bias 

mixers in radio receivers), control of the amplitude by bottoming is 
usually objectionable. 

Also, the bottomed oscillator is not capable of giving very low 
distortion, and may be unsuitable for some applications. 

For such circuits, the mean current has to be well defined at a value 
less than that which causes bottoming. A suitable bias arrangement 
is the conventional potential-divider and emitter-resistor circuit shown 
in Fig. 8. The potential divider should normally pass a current which is 
about ten times the expected base current of the transistor. 

For transistor mean currents up to about 5mA, there is little difference 
between the mean current and the direct or quiescent current when not 
oscillating, so that the values of R b i, Rb2 and R e can be determined 
as for a small-signal amplifier. Also, the effect of Rf on the bias network 
can be neglected. Thus up to 5mA, 




Above 5mA, the oscillatory current is somewhat greater, and if 
complete accuracy is required, the bias voltage (Fig. 3) must be added 
to Vbb before calculating lmean . 

Page 227 


Hence above 5mA, 


lmean — 


Furthermore, in the circuit of Fig. 8, if too small a value is chosen 
for Vbb , the change in the bias voltage with the variation in loop gain 
from transistor to transistor will cause a large change in Imean and there- 
fore in output voltage. 

Vbb may be obtained from a tapped battery in the usual way, if 


The temperature stability of the mean current of this circuit is similar 
to that with a small-signal amplifier. For instance, the change in 
Vbe ( ~ — 2mV/°C) has the same effect upon the mean current as the 
same change in Vbb. A 10% change in Vbb , for example, will cause a 
10% change in mean current. Vbb can be chosen sufficiently high to 
keep the variation in mean current within the limit required for any 
particular application. 


Since the above types of oscillator supply current pulses to the tuned 
circuit, the oscillators usually work in class C, and the distortion is 
greater than for class A operation. 

The distortion can be limited as desired by a suitable choice of the 
working Q (Q w ), as the following example illustrates. For a loop gain 
A of 3 or 4 ,the second-harmonic current is typically equal to 0-6 
times the fundamental. Then the collector voltage will contain 1% 
second-harmonic distortion when 

0-6 X 

(— )= — 
\3Qw/ 100 

whence Q w = 40. 

Also, the third-harmonic current is typically 0-25 times the fundamental. 
The collector voltage will contain 1 % third-harmonic distortion when 

G - 25x (s^) = i5o > 

that is, when 

The factors 2/3Q w and 3/8Q w are obtained from considerations of 
the impedance of the tuned circuit to the respective harmonics and 
therefore of the voltage that will be developed. 

Page 228 



The value of C in the circuit of Fig. 8 has to be chosen carefully in 
order to avoid squegging. As a general rule, the reactance of C should 
be about re to | of the value of (r e +Rf). The situation is aggravated 
by large amounts of feedback, and A should be restricted to a maxi- 
mum of about 4. The position is worst for low-power stages, where the 
high value of R e gives a large value to the time-constant CxR e . 

Squegging is much less of a problem with the circuit of Fig. 7, when 
this has an amount of bottoming, but it is good practice to use a capacit- 
ance which only just gives the necessary decoupling. A useful guide is 

a'min(r e + Rf) 





Each circuit should be designed to cope with the variations in tempera- 
ture and supply voltage present in the particular design. 

The inductance of the coil used in the tuned circuit is a function of 
temperature and, for a magnetic core, it is also a function of the peak 
output voltage. The Ferroxcube pot core for which these effects are 
least troublesome is the LA6. 
For the LA6, 

SL ~ +6 parts per million 
for every 1 % increase in peak output voltage. Also 

8L < +60 parts per million 
for every 1°C rise in ambient temperature. 
For an air-cored coil, a variation of 

SL ~ +25 parts per million 
per C C rise in ambient temperature is typical. 

In the transistor itself, the principal frequency drift is caused by the 
phase shift introduced by the internal base resistance. The phase 
shift cj> of vin referred to the feedback voltage e s is 

<(> = tan" 1 (rbb'/Xce) if X ce > r b b- . 


<[> = tan -1 


r e fi 

where f r is the resonant frequency of the tuned circuit, and is approxi- 
mately equal to the operating frequency, and fi is the high-frequency 
parameter. Provided <j> is less than 0-5 radian, 

Page 229 


The phase shift between voltage and current in the tuned circuit 
is given by 

V = 2Q W A 

a f r- f 
where A = — 7— 


and f is the operating frequency. 
The phase shift <j> is balanced out by <j>', so that <j>' = — <j>, whence 

. fbb'fr 

~~ 2Q w r e fi ' 
As an example, suppose r b b- = 100Q, f r = lOOOc/s, fi = 500kc/s, 
Q w = 10, and r e = 25Q. Then A = -400 x lO" 6 . For a 10% decrease 
in r e , caused by an increase in collector current with temperature, 
A decreases by 40 parts per million. 

20mW, lOkc/s OSCILLATOR 

As an illustration of the design procedure, a requirement is postulated 
for a lOkc/s grounded-emitter oscillator. The supply voltage is to be 
12V and the peak output power about 20mW (that is, lOmW mean) 
at 20°C, and this output must not increase by more than 10% if the 
ambient temperature rises to 45°C. Further, the oscillator must not 

D.C. Stabilisation 

First the stage is stabilised to have a quiescent current which does not 
increase by more than 10%. The stability calculations are essentially 
the same as those for a class A amplifier stage with a mean current 

Imean • 

Briefly, the 25°C increase in temperature causes a 50mV decrease in 
the base-emitter voltage V be for the same quiescent current, so that the 
base supply voltage Vbb must be at least 0-5V, if the quiescent current 
is not to increase by more than 10%. Vbb may be obtained from a 
potential divider, provided the source resistance is sufficiently low for 
(a) Ico not to cause trouble, arid (6) for the mean base current (which is 
dependent upon a' — a quantity which differs from transistor to transis- 
tor) not to introduce any appreciable change in Vbb . The second 
condition is satisfied if the potential divider has a standing current of 
about ten or more times the base current. 

As high efficiency is not a requirement, a base supply voltage Vbb of 
1 -5V may be used. The temperature dependence of the oscillator bias 

Page 230 


may be ignored, because it is insignificant compared with the change 
in V be . 

Load Resistance 

If the collector voltage is allowed to swing within 2-5V of the emitter 
voltage, the peak oscillatory voltage will be 

12-1-5-2-5 = 8V, 
where the voltage dropped across the emitter resistor is approximately 
equal to Vt, b ( — 1-5V). The amplitude will not be sufficient to bottom 
the transistor when it increases by 10% at the higher temperature. 
The collector load required to dissipate a peak power of 20mW is 

R _ _ (Vout(pk)) 2 82 

RL ~ ~T^T = 20*10* = 3 ' 2ka 

Mean Stage Current 

Because the peak collector current at the fundamental frequency is 
approximately twice the mean current, 

T _ Vout(pk) 8 

I mean _ _____ _ __ _ ! . 25mA . 

Emitter Resistor and Base Potential Divider 

The value of R e can now be calculated from 

t> Vbb— Vbe+Vbias 

Re= = • 


Vbias can be obtained from Fig. 4, and is 140mV for a loop gain of 
say, 4. Thus 

Re = -- = M9_Q _ l-2k__, 

where 0- 15V is a typical figure for the bias voltage of a low-power 
germanium transistor operating at an emitter current of l-25mA, and 
l-2kQ is taken as the nearest standard resistance value. 

The resistances in the potential divider are calculated in the usual way. 
Transformer Ratio 

The loop gain (or more precisely the initial loop gain) which has been 
chosen to be 4, is given by: 

A r l 
Ao = gmo X ' 

where g mo is the value of g m immediately before the onset of oscilla- 
tion, and is the small-signal g m with the direct current of the transistor 
equal to the mean current of the oscillator. 

Page 231 


The value of g mo in A/V is equal to I m ean/25 if Imean is in mA, so 


1-25 3200 
25 X n 

from which n = 40. The source resistance looking into the secondary 
of the transformer is 

R L /n 2 = 3200/40 2 = 2D. 
This value justifies the assumption that the transistor may be considered 
as being driven from a voltage source. 

Complete Circuit 

The complete circuit is shown in Fig. 9. A working Q of 10 is used (the 
unloaded Q being very much greater than 10), so that the collector 
load can be specified as nearly as possible by an ordinary resistor. 

If the output voltage is calculated more exactly, it is found that, 
with an A of 4, the approximation 



gives an output voltage which is 7% too high (Fig. 4), so that the 
expected peak output voltage is changed from 8 to 7 -44V. In practice 5 

Fig. 9 — 10kc/s oscillator for output of 20mW 

the mean current would be increased by 7 %, but this adjustment has 
not been made to the circuit of Fig. 9. 

A loop gain greater than 4 should not be used, because complete 
decoupling of the emitter is then not always possible. The partially 
unbypassed emitter resistance causes degeneration; this reduces the 
effective loop gain below the calculated value, and the output voltage 
may be more difficult to calculate because of the phase shift. Because 
of the high degree of feedback, the circuit will oscillate with a supply 
voltage of down to about 3V. 

Page 232 



The 3(xF capacitor decoupling the emitter has a reactance of 5Q at 
lOkc/s and represents the largest practical value that can be used with- 
out squegging, which starts at 4[xF. This capacitance needs to be chosen 
carefully if squegging is to be avoided, but in the design described, 
in which the loop gain does not change appreciably from transistor to 
transistor, the problem is not a difficult one. This component should 
be chosen with care; if necessary three lfxF paper capacitors should be 
used in parallel. The resulting capacitance at the maximum of the 
tolerance range should not exceed 4(jlF. 

90mW, 50kc/s OSCILLATOR 

The design procedure will not be given for this oscillator (Fig. 10), 
as the requirements for this circuit were deliberately made as difficult as 

14 ,,18 5mA(mean) 
^ ^__ 

> 435 [ 88 V 
> n | peak 

_ — " X - 

Fig. 10 — 50kc/s oscillator for output of 90mW 

possible whereas, in general, the design problems would be much easier. 
The circuit has been designed to work with the 'worst' transistor. 

The oscillator operates at 50kc/s from a 12V supply, and provides 
an output power of about 90mW. The circuit is arranged so that all 
transistors will be bottomed. When fifty OC72 were tried in the circuit, 
the output voltage showed a spread of only 2 %. 

Care should be taken with the base-resistor-bias circuit — such as 
that shown in Fig. 10 — to ensure that the oscillator output is never 
short-circuited and that the polarity of the feedback winding is never 
reversed. Otherwise a' is fully effective with no oscillation, and the col- 
lector current might be destructively high with high-a' transistors. 

Because of the stabilising action of the 47Q unbypassed emitter 
resistor, the circuit will oscillate with a supply voltage of down to 
about 4V. 

Page 233 



The oscillator shown in Fig. 1 1 has been included to help those who 
are experimenting with transistor circuits for portable tape recorders. 
A single-transistor oscillator is sufficient to provide the small amount 
of power required for a.c. biasing. Considerably more power would 
be needed to cover the requirements of a.c. erase, but this function 
can be performed by a permanent-magnet system. 

The bias oscillator of Fig. 11 is based on the circuit given in Fig. 10. 
The frequency of oscillation of 42kc/s has been chosen as being five 
times the highest frequency to be reproduced in a tape recorder of 
reasonable quality. The circuit will operate satisfactorily with any 
OC72 and will continue to oscillate with a battery voltage of only about 

820pF J 

I50turns Record/ 

34 ^f~N playback 
s.w.g. <E j head 

Fig. 11 — 42kc/s oscillator for feeding playback/record head of tape recorder 

one-third of the nominal value. The current consumption is 5 to 10mA 
at 9V. The output voltage of between 32 and 35V peak is sufficient to 
provide a minimum of 1mA bias current in the 250mH inductance of 
the record/playback head shown in the circuit. 

Page 234 



The design of RC phase-shift oscillators is similar for transistors and 
thermionic valves, but three factors must be borne in mind : 

(a) the RC phase-shifting network is required to feed into the input 
of the transistor, which is of low impedance 

(b) the internal phase-shift of the transistor will be added to, or 
subtracted from, that of the network 

(c) when ladder networks are used, the current amplification factor 
of the transistor must be appreciably greater than the attenuation 
of the network. 


Where it is required to produce an oscillator with only one transistor, 
the grounded-emitter configuration will be chosen, because of its 
high gain. From the small-signal grounded-emitter equivalent circuit 

"I e 

- = 500 to 5000 typically 

Fig. 1 — Small-signal grounded-emitter equivalent circuit 

shown in Fig. 1, it may be seen that the output-current generator 
g m v is in antiphase with the voltage v derived from the input voltage vi n . 
A network giving a phase shift of 180° between the output and input 
of the transistor is therefore wanted. Ladder networks are commonly 
used for this purpose. 

Ladder Networks 

Ladder networks may be built up by cascading a number of similar 
RC phase-shifting sections, using any one of the sections shown in Fig. 2 
as a basic element. 

Sections (a) and (b) give a phase shift between the input and output 
currents, so that the input of the transistor, which is connected across 

Page 235 


the pair of terminals on the right-hand side, is fed with a current. The 
input impedance of the transistor should preferably be much less than 
the impedance of the network — which is R in (a) or the reactance of 
C in (b) — in order that the transistor impedance shall not disturb the 
operation of the circuit. This requirement is fairly easy to meet. 
However, an a' of at least 60 is needed for satisfactory operation, and 


R 'out 'in C 'out 
-WW — ) o o > 1| ) — 



HI i 

T i 

Fig. 2 — Basic elements of ladder networks 

since the input impedance of the transistor in grounded-emitter con- 
nection may be as high as 2-5kiQ or more at I e = 1mA, the resulting 
circuits are of high impedance, and a collector supply voltage of 12V 
or more is needed. 

Sections (c) and (d) in Fig. 2 are voltage-transfer networks, and are 
normally employed with thermionic valves. For transistors, the input 
impedance of the transistor, which is connected across the right-hand 
side, needs to be large compared with the impedance of the network. 

The voltage vi n across the left-hand terminals of (d) could be generated 
by allowing the output-current generator of the transistor (Fig. 1) to 
work into R, as may be seen from the equivalence in Fig. 3. No such 

where e = iR 

Fig. 3 — Equivalence of i, R arrangement to e, R arrangement 

simple method can be found for (c), and this circuit may be dismissed 
as unsuitable. 

This way of generating vm sets a lower limit to the value of R which 
may be employed while obtaining sufficient voltage for oscillation. 

Page 236 


The condition is R > Ar e , where A is the attenuation factor of the 
network, and is equal to 29 for a network of three equal sections. 
At I e = 1mA, r e is equal to 25/I e = 250, and R needs to be greater 
than 29 x 25 = 725Q (say, l-2kQ). 

To have an input impedance which is sufficiently high for satisfactory 
operation, the transistor needs to have an a' greater than 100. These 
networks therefore are better suited to thermionic valves than to tran- 
sistors; although, if transistors of sufficiently high a' are available, 
the circuit can be made to work from lower supply voltages than 
when using sections (a) or (b). 

Number of Sections 

The most suitable RC phase-shift networks may be built up, therefore, 
from sections of the (a) or (b) type. 

Neglecting the internal phase shift of the transistor for the moment, 
at least three sections are necessary, and the networks could be of 

R R R 

vVv^- t-^V\AA-f-vW\A 


f-VWV*- t■^^AAA-f-^ 

_IL IU' 

r> T-Hh t II H H 

< a < c 


Fig. A — Preferred ladder networks for transistors 

the form of (a) or (b) in Fig. 4. If circuit (b) is used, the first R can be 
the load resistance of the transistor, and the circuit design is consider- 
ably simplified. 

The current attenuation of these networks, at the frequency where 
the phase shift between the input and output currents is 180°, is 29. 
To allow for losses in the input and output impedances of the transistor, 
the a' should be greater than about 60. The OC75 is therefore suitable, 
though transistors at the top end of the OC71 production spread 
should also work in this type of circuit. 

The current attenuation of 29 applies to a network having three 
equal resistances and three equal capacitances. These are the simplest 
to design and the most commonly used. Also, since the grounded- 
emitter transistor has an input impedance which is only an order or so 
different from its output impedance, tapered networks are of little use. 

Operating Frequency 

Single-transistor phase-shift oscillators are best restricted to low- 
Page 237 


frequency operation, where the internal phase shift of the transistor 
need not be considered. At higher frequencies, both the phase shift 
and the reduction in a' cause design difficulties. For instance, with 
the network shown at (b) in Fig. 4, the phase shifts in the network and 
in the transistor vary in opposite senses, so that at high frequencies 
more phase shift has to be provided by the network. The phase shift 
in each section has to be increased, or a fourth section added. There 
is consequently more attenuation. 

With network (a), the transistor and network phase shifts vary in 
the same sense, and a two-section oscillator can be constructed, though 
the operating frequency will be somewhat dependent upon the particu- 
lar transistor. Also, since any shunt resistance across the input capacity 
of this network reduces the phase shift, a higher collector-load resistance, 
and possibly a higher transistor output impedance, will be demanded. 

From Fig. 1, it may be seen that when the transistor is current fed, 
the phase shift in v, and hence in the output-current generator, will 
be 45° when the reactance of c e is equal to a' r e . This frequency, which 
is above that which would normally be used with circuit (b), is called 
the grounded-emitter cut-off frequency, and is designated by f' a . It is 
given by f' a = fi/a', where fi = f a /l-22 for alloy-junction transistors. 

800c/s Oscillator 

A circuit using the (b) network and operating at 800c/s is shown in 
Fig. 5. Ideally the operating frequency is given by f = l/27rCRV6, 

Fig. 5 — 800c/s oscillator using three-section ladder network 

which for the circuit shown is 650c/s. The transistor input and output 
impedances modify this to 800c/s. 

The value of the phase-shift resistors (lOkQ) is chosen to be a mean 

Page 238 


between that which will be appreciably affected by the transistor 
output impedance, which is high, and that which will be appreciably 
affected by the transistor input impedance, which is low. With these 
networks, it is not easy to control the amplitude of oscillation without 
somewhat affecting the frequency of operation, and the amplitude 
control may change the frequency by 10 % or so. The gain is controlled 
by changing the distribution of the feedback current between the base- 
bias resistors and the transistor input, the unbypassed resistance in 
the emitter increasing the transistor input impedance. The control 
should be adjusted so that the oscillation amplitude is smaller than 
that giving objectionable distortion. 


RC oscillators may also be constructed by arranging the phase-shift 
components to form a Wien network. The advantage of this network 

'0/OC71 ^^ oc 


Fig. 6 — Two-transistor Wien-network oscillator for about 3-4kc/s 

is that the attenuation factor is only 3 at the frequency which gives 
zero phase shift although, since the output is in phase with the input, 
at least two amplifier stages are necessary. 

Two-transistor Circuit 

A simple two-stage oscillator of this type is given in Fig. 6, where 

Page 239 


the Wien network is R7, C3, C4, R9. Both stages are d.c. stabilised, 
the lower base-bias resistance of Trl also being part of the bridge 

Fig. 7 — Three-transistor Wien-network oscillator for 15c/s to 20kc/s 

network (R9). The 3-3kO emitter resistor (R8) of Tr2 is left unbypassed, 
so that this stage has considerable a.c. negative feedback. 

RV3 in the emitter of Trl provides a convenient means of adjusting 
the waveform for amplitude and distortion, and of compensating for 
changes in temperature from, say, one day to the next. The waveform 
is good, and the short-term temperature stability is fairly good. Supply 
voltages from at least — 3V to — 6V may be used, provided RV3 is 
suitably adjusted. The output may be taken from the collector or 
emitter of Tr2 (R7 or R8). 

This circuit operates very nearly at the theoretical frequency of 

1 1 

27rV(C 3 C 4 R7R9) ~ 27rC 3 R 7 ' 
since in fact C3 = C4 and R7 = R9 . R7 can be shown to be effectively 
in series with C3 by means of the equivalence given in Fig. 3. 

Page 240 


With the capacitances shown, the circuit oscillates at a frequency 
of about 3-4kc/s. For operation at other frequencies, the capacitances 
should all be increased or decreased by the appropriate factor. (The 
resistances cannot be changed without altering the d.c. conditions.) 
The values of C3 and C4 should be chosen according to the frequency 
accuracy required, but for CI and C2, the nearest standard values 
may be taken. 

The circuit is sufficiently uncritical of gain to accept the OC70 or 
OC71, but if operation is required at higher frequencies, it will probably 
be better to use two OC45. 

Three-transistor Circuit 

A more professional Wien-bridge oscillator is shown in Fig. 7. This 
circuit incorporates a thermistor R5 as an amplitude-control device, 
and the output is essentially independent of small changes in supply 
voltage or ambient temperature. (A suitable component for R5 is the 
S.T.C. thermistor type R53.) Apart from the frequency-determining 
capacitors, only one capacitor is required. Consequently the unit is 
compact, and no difficulties arise from phase shift in the coupling 

The output voltage is IV r.m.s., and the circuit operates with supply 
voltages between 7V and 12V and consumes about 10mA. In Fig. 7, 
a lower limit of 9V has been set to the supply voltage to ensure low 

The frequency coverage is from 15c/s to 20kc/s in three ranges: 
15 to 200c/s, 150 to 2000c/s and 1-5 to 20kc/s, the lower frequencies 
being associated with the larger capacitances. The ganged variable 
resistors, RV9 and RV11, allow the frequency to be adjusted within 
any given range. In an experimental model, the amplitude over the 
full range was constant to within better than 2%. 

If it is desired to extend appreciably the upper frequency limit of 
the oscillator, Tr3 should be changed to an OC41. This modification 
will enable the oscillator to work satisfactorily at frequencies in excess 
of lOOkc/s. For other transistors, R8, the 6-8kQ bias resistor feeding 
the bridge, may need to be adjusted to ensure optimum working points 
for the transistors. 

The OC140 (Tr2) is an n-p-n transistor, and its connections should 
therefore be made as in the circuit diagram, with the emitter connected 
to the negative supply line. 

To avoid excessive distortion, the external load connected to the 
oscillator when the output is at its maximum should be not less than 
lk£l. With the addition of the load, the change in the maximum output 
voltage is less than 1 %. 

Page 241 



With a change of 3 V in the supply voltage (from 9V to 12V), the change 
in frequency at lOkc/s and the change in output voltage are less 
than 1 %. 

Page 242 



Many transistor pulse circuits are similar in form to classic thermionic- 
valve circuits, and perform similar operations. However, the transistor 
gives a closer approach to the ideal switch than the thermionic valve, 
and the limitations of the two types of device are different. The design 
of transistor circuits is based, therefore, not upon analogues, but on 
an analysis of the operating states of the transistor. Most of this chapter 
is consequently given over to the large-signal equivalent circuit. 

In most pulse circuits, the transistor is used as a voltage or current 
switch, the timing operations being performed by combinations of 
resistance and capacitance (CxR), or resistance and inductance (L/R). 
The limit to the pulse repetition frequency is set ultimately by the switch- 
ing times of the transistor. 


The basic circuit of the transistor as a voltage switch is given in Fig. 1. 

On Condition 

The collector current that the transistor is capable of passing in the 
normal condition is al e ( = aT b ). If the collector current is limited 
to some lower value— for example, by a resistive load— the emitter 
and base currents are higher than those required for normal transistor 



Fig. 1 — Transistor as voltage switch 
Fig. 2 — Transistor as current switch 

action. This is the bottomed (or current-saturated) condition. Both 
the collector and emitter diodes are forward biased. 

Under normal operating conditions, the holes diffusing across the 
base of the transistor are swept rapidly across the collector depletion 
layer. As the collector current increases (Pig. 1), the voltage across 
the transistor is reduced. Consequently the field across the depletion 

Page 243 


layer is diminished. To the first order of approximation, however, the 
flow of holes across the collector junction is not affected by the collector 
voltage. Even in the bottomed state, when many more holes leave the 
emitter than are required to maintain the maximum load current in 
the collector, the normal flow of holes across the collector junction is 
not affected. The excess holes that cross the junction bias it in the for- 
ward direction, so that these holes recross the junction into the base 
as a forward current through the collector diode. 

The bottoming voltage in common (or grounded) emitter is there- 
fore the difference between the voltages across the two forward-biased 
diodes. This voltage is quite small — about lOOmV for alloy-junction 
transistors — and smaller than the knee voltage usually considered in 
the design of a.f. amplifiers. 

Off Condition 

If the base-emitter voltage is greater than I co rbb' , the emitter junction 
will be reverse biased. In this condition, therefore, both the diodes are 
reverse biased. The collector and emitter leakage currents are then 
given (as will be shown later) by: 

, Ico(l—ai) , u 

lco(o) = -, • • -\\) 

1 — <XN«I 


leo(o) — 

Ieo(l — «n) n) 

1 — a^oci 

where <xn is the normal large-signal a of the transistor, and ai is the 
large-signal a with the collector used as an emitter and the emitter 
used as a collector (inverse connection). 

A suitable step of current on the base of the transistor will switch 
the collector voltage between the bottoming voltage and a voltage 
nearly equal to the supply voltage V cc - 


The basic circuit of the transistor as a current switch is given in Fig. 2. 

On Condition 

Assuming that the emitter supply voltage V ee is much greater than the 
switching voltage applied to the base, then a current 

~ Re 

is switched into the emitter of the transistor when the base is made 
more negative than the voltage dropped across the emitter junction. 

Page 244 


(V be is not more than 0-3V for the OC41 and OC42 at I e = 10mA.) 
The collector current in this condition is 

~ aV ee 


When the transistor is used as a current switch, it must not be allowed 
to bottom; otherwise the output impedance would become very low, 
and the transistor would cease to act as a constant-current generator. 

Off Condition 

When the base is made positive to ground by a voltage greater than 
the voltage drop across the emitter junction, the emitter junction will 
be reverse biased, and a current 

~ Re 

will be switched into the diode. (In practice the diode may be the 
emitter diode of another transistor.) The collector current in this 
condition is I C o(o) • 

Thus a voltage step of about IV on the base of the transistor will 
switch the collector current from I CO (o) to approximately aV ee /Re. 


The analysis of the operating states leads to the setting up of a large- 
signal equivalent circuit, from which the switching times of the transistor 
can be calculated. As this approach will be extensively employed in 
future, the derivation will be given in full. 

The equivalent circuit for large-signal operation may be derived from 
a simple physical model. It is assumed that the transistor is a p-n-p 
type, but by reversing the sign where appropriate, the derivation will 
apply equally well to n-p-n transistors. Only alloy-junction transistors 
will be considered. 

D.C. Equivalent Circuit 

The large-signal d.c. equivalent circuit of an alloy-junction transistor 
is shown in Fig. 3. The diodes D e and D c represent the emitter and 
collector junctions of the transistor. As the emitter and collector 
junctions are very close together and form a single crystal with the 
base, part of the current flowing in at one junction will flow out at the 
other. These currents are represented on the equivalent circuit by the 
current generators aNiei and aii c i (where n and i denote normal and 
inverse connection). An appreciable resistance exists between the 
external base contact and the active area of the transistor. This resist- 
ance is represented in the equivalent circuit by rbb' . The resistivity of 
the emitter and collector material is lower than that of the base, and 

Page 245 


the currents in the emitter and collector flow through regions of much 
greater cross-sectional area. The internal resistances in series with the 
emitter and collector are therefore negligibly small, and are not 
included on the equivalent circuit. 

Depletion Capacitance 

When the voltage across either junction of the transistor changes, the 

Fig. 3 — D.C. equivalent circuit for large signals 

width of the corresponding depletion layer also changes, and a 
charging/discharging current must flow. To take this effect into account, 
the depletion layers are represented by capacitances c e (dep) and c C (dep) 
on the full equivalent circuit (Fig. 4). The depletion capacitance of 



~" !&+ 

Fig. 4 — Full equivalent circuit 

alloy-junction transistors is inversely proportional to the square root 
of the voltage across the junction. 

Diffusion Capacitance 

The flow of current carriers through the base layer arises from the 
diffusion of holes within the field-free region of the base. There will 

Page 246 


therefore be a concentration gradient of holes in the base, proportional 
to the flow of current through the base. 

The hole-concentration pattern for a transistor operating in the 
normal active region is shown in Fig. 5(a). In the absence of a drift 
field, the concentration gradient is uniform and is represented by a 
straight line. The hole concentration at the collector junction is zero, 
because all holes which reach the junction are swept into the collector 

Concentration gradient 
for increased emitter 

Increased charge 
required to alter the 
concentration gradient 

(a) Normal regit 

(c) Bottomed region 

Fig. 5 — Distribution of base charge for (a) normal region of characteristics 
(fa) inverse region (c) bottomed region. 

by the potential gradient which exists across the collector depletion 
layer. As can be seen from Fig. 5(a), when the emitter current changes, 
a capacitive current must flow to alter the hole-concentration gradient. 
The value of the capacitive current is given by 

1 diel /<i\ 

ie2 = -7- > ■'•{*) 

coi(N) at 

where g>i(n) is the angular frequency at which the normal a' becomes 
equal to one. 

Also, it can be shown that the charge required in the base to provide 
a collector current of a^Ie is given by 

Q b = -^-- -..(4) 


When the transistor is operating in the inverse active region, the 
concentration pattern is as shown in Fig. 5(b). The only difference 
between this and the previous case is that, since current is flowing from 
collector to emitter, the concentration gradient is reversed. By similar 
reasoning to that used to derive Eq. 3, it can be shown that the capaci- 
tive current is given by 

1 did (*\ 

1 C 2 = -tt' 'W 

<oi(i) dt 

where g>kd is the angular frequency at which the inverse a' becomes 
equal to one. 

The effect of the charges in the base can be taken into account by 
including two capacitive current generators i e 2 and i C 2 in the full 

Page 247 


equivalent circuit (Fig. 4). The values of i e <2 and i C 2 are given by 
Eqs. 3 and 5. 

When the transistor is operated in the saturation region, the con- 
centration gradient is given by the line xy in Fig. 5(c). The concentra- 
tions given by this line can be considered, to be the sum of the con- 
centrations given by the lines xc and ye, which are represented on the 
equivalent circuit by the two capacitive generators i e 2 and i C 2- 

Simplification of Full Equivalent Circuit 

The diodes D e and D c of Fig. 4 have characteristics of the form shown 
in Fig. 6(a). When the junction is reverse biased it can be considered 

= 's~p2t-' 

Fig. 6 — Complete diode characteristic (a) and simplification of characteristic (b) 

to be a current generator I s . The equation of the curve is then given by 
I-I s {exp(qV/kT)-l}. ...(6) 

A linear approximation may be made to this characteristic, in order 
to simplify the mathematical expressions obtained when using the 
equivalent circuit. 

The linear approximation is shown in Fig. 6(b). When the junction 
is biased in the forward direction, the diode is replaced by a resistance 
r in series with a cell V . The value of r which correlates best with 
practical results is given by the slope of the diode characteristic at the 
peak current. The value can be shown, by differentiating Eq. 6, to be: 

r = 



The value of V is given by the intercept of the tangent at any particular 
value of ipt on the voltage axis. 


The equivalent circuit for the cut-off state (Fig. 7) can be derived from 
the full equivalent circuit given in Fig. 4. Since D e and D c are reverse 
biased, they can be represented by current generators I es and I cs . The 
capacitive currents i e 2 and i C 2 are zero, because they are proportional 

Page 248 


to the rates of change of I es and I cs , which are constant. The current 
generators aidei and aii c i in Fig. 4 become aNles and ajI C s in Fig. 7, 
since i e i = —Ies and i c i = —Ics when the junctions are reverse biased. 
I es and Ics are the saturation currents of the emitter and collector 
diodes with the opposite diode short-circuited. These currents can be 


e dep. 

: c dep. 


Fig. 7 — Equivalent circuit for cut-off region 

expressed in terms of I eo and I co . From Fig. 7, the value of the 
collector current with the emitter open-circuited is 

Ico = lcs _-a Nles 5 

and since in this state 

Ies — (Xllcs » 
Ico = Ics(l — aN«i) . 





IfiS — 

1 — <XNai 



1 — aN<xi 

The value of collector current when the emitter junction is reverse 
biased is obtained from Fig. 7, thus : 

Ico(o) — Ics <*Nles > 

and since now 

a^Ies = ail cs , 
Ico(o) = Ics(l — ai) 

Ico(l— ai) 



1 — awai 

_ Ieo(l — «n) 
1 — a^ai 

Page 249 

(from Eq. 8). 



In the normal active region the emitter diode D e (Fig. 4) is biased in 
the forward direction. It can therefore be represented by a resistance 
r oe in series with a cell V oe (Fig. 8). The capacitive current 

1 diei 

le2 = • -T- 

(Oi ( N ) dt 

will then flow through a capacitance of 


C e = 

<*>l(N)r e 

in parallel with r oe , since 

dV 1 d(ieir oe ) 

le2 = Ce ' -j- = -j- 

at coi(N)roe dt 


1 diei 

0)1 (N) dt 

The depletion capacitance c e (dep) is not included in Fig. 8, because it 

f- 1/ U)1(N)i-oe 

Fig. 8 — Equivalent circuit for normal region 

is negligibly small in comparison with the diffusion capacitance c e . 
This equivalent circuit is used to calculate switching times. The 
current generators (I C s and ajl cs ) have very little effect on switching 
times and have therefore not been included in Fig. 8. 


The considerations leading to the derivation of the equivalent circuit 
for the normal active region apply equally well to the inverse active 
region and result in the circuit of Fig. 9. Here, however, the inverse 
parameters ai and taio are used instead of the normal parameters. 


Fig. 10 shows the equivalent circuit for the bottomed region. The 
diodes D e and D c (Fig. 4) are biased in the forward direction, and are 
represented by resistances r oe and -r oc in series with the cells V oe and 
V c- The capacitive currents i e 2 and i C 2 flow through the diffusion 

Page 250 


capacitances c e and c c . The depletion capacitances c e (dep) and c c «iep) 
are omitted because they are negligibly small in comparison with 
c e and c c . 

Use of Large-signal Equivalent Circuit 

The equivalent circuit derived above can be used to calculate the 
behaviour of the transistor under large-signal conditions. In particular 

Fig. 9 — Equivalent circuit for inverse region 

the switching times of the transistor may be calculated. The switching 
times are evaluated by forming a differential equation for the equiva- 
lent circuit of the transistor and the external circuit, and then solving 
the equation. The theoretical values thus obtained agree reasonably 
well with those measured in practice. 


The equations governing the performance for generalised conditions are 
quite complicated. As an example of the use of the large-signal 
equivalent circuit, it will therefore be convenient to consider only a 

Fig. 10 — Equivalent circuit for bottomed region 

special case, that in which the transistors are overdriven (as is common 
practice) to obtain fast switching times. 

Recombination can be assumed to give an approximately exponential 
decay of base charge, and a recombination time-constant may be 
defined from the exponential curve in the usual way. If the switching 
time is small in comparison with the recombination time-constant, the 

Page 251 


effect of recombination may be ignored. The switching time can then 
be found by dividing the required base charge by the base drive 
current, since by definition [Q] = [t] x [I]. 

The recombination time-constant can be derived as follows : 

From Eq. 4, 

o Ie 


The recombination current is given by 

T Ie 

assuming an emitter efficiency of one. The recombination time-constant 
is therefore given by 

Qb = 1+5' . 

lb <*>1(N) 

The base charge required to switch the transistor on to a collector 
current I c can be found from the equivalent circuit (Fig. 8). If recom- 
bination is ignored, ocn = 1 and all the drive current is available to 
charge c e and c C (dep) • The turn-on time is then given by 

ton== 9^£) ...(11) 

lb (on) 

and the charge required to switch the transistor on is 

Qb(on) — Qce+Qc C( dep) 



Cc(clep)AV c •••(12) 


where AV C is the change in collector voltage. 

The charge required to switch the transistor off, if it is not bottomed, 
is equal to that required to switch it on. When the transistor is bottomed, 
additional charge is required to switch it off, in order to overcome the 
excess base current. The extra charge is 


T S 

where t s is the hole-storage time-constant of the transistor and is 
given by 


T S 

1 — a^ai 
The total charge required to switch the transistor off is then 

Page 252 



Qb(off) — - 


The turn-off time is given by 

C C (dep)AV c + 


T S . 

toff = 

lb (off) 




In relay-switching circuits, the transistor operates essentially as a 
power switch. A typical relay drive circuit is shown in Fig. 11. R L is 
the resistance of the relay winding. 


Fig. 11 — Relay drive circuit 

On Condition 

When Vm is negative, the negative base current (— Ibi) flowing out of 
Tr2 is sufficient to bottom this transistor. The current in the load rises 
to V CC 2/Rl . The total dissipation in the transistor is given by 

V C eVcc2 



-V be Ib • 

The power in the load is (V CC 2) 2 /Rl and is comparatively high. 

Off Condition 

When V ln goes positive, Trl is cut off. A positive base current of 
approximately +I b2 = V b b/(Rb+r b b') flows into Tr2, bringing Tr2 
out of bottoming and tending to switch it off. 

The collector current in this condition (emitter and collector diodes 
reverse biased) is I CO (o) . The dissipation in the transistor is approxi- 
mately V CC 2lco(o) and the power in the load is (I CO (o)) 2 Rl . 

Delayed Switch-off 

A description of the delayed switch-off which occurs with an inductive 
load has already been given in Chapter 9 (pp. 101 and 102). 

Page 253 


The induced voltage at switch-off is usually sufficient to make the 
collector potential rise almost immediately to the supply voltage. 
The relay should be shunted by a catching diode, such as Dl in Fig. 11, 
which will conduct when V c = V CC 2 and prevent any further rise in 
collector voltage. The OA81, OA5 or OA10 should be used, according 
to the peak current and the voltage to be handled. 

(V CC 2 < V X ) V x (V CC2 >V X ) V ce 

Fig. 12 — Static output characteristic illustrating switch-off loci 

During switch-off, the operating point (Fig. 12) follows some path 
such as ABC (V CC 2 < V x ) or ADEF (V CC 2 > V x ). The transistor 
is then in a high-dissipation region. The switch-off time may be only 
a few microseconds for the path ABC, but may be tens of milliseconds 
( - L/Rl) for ADEF. 

Overheating of the transistor for the condition V CC 2 > V x can be 
prevented by shunting the relay by a capacitance C (Fig. 11). 
By choosing a suitable value for C, the switch-off time can be shortened 
and the locus of the operating point modified to some path, such as 
AGF, where the dissipation is lower. 

The value of C which ensures that the collector current reaches zero 
by the time the collector voltage reaches some value Vi is given by 
C = It/2Vi , where I is the collector current immediately before 
switch off and t is the maximum switch-off time of collector current 
(assumed linear). C should not be given a larger value than necessary, 
since it increases the dissipation during switch on. 

Page 254 



Switching between two operating states may be achieved by means of 
regenerative feedback. The two states may both be unstable (multi- 
vibrator or astable circuit) or one may be unstable and the other stable 
(flip-flop or monostable circuit), or both may be stable (Eccles- Jordan 
or bistable circuit). The family resemblance between the three types of 
circuit is brought out by the basic circuits of Figs.l, 2 and 3. The distin- 
guishing features of the circuits are the cross-coupling components. 

MULTIVIBRATOR (Astable Circuit) 

The multivibrator (Fig. 1) will be considered first, as this is probably 
the best known and, in appearance at least, the simplest. The circuit in 

Fig. 1 — Basic multivibrator (astable circuit) 
Fig. 2 — Basic flip-flop (monostable circuit) 
Fig. 3 — Basic Eccles- Jordan (bistable) circuit 

its thermionic-valve form was described by Abraham and Bloch in 1918. 
It was first conceived as a square-wave generator, and since a square- 
wave is very rich in harmonics, the name multivibrator was .coined. 
Upon switching on the circuit, slight unbalance in the components, 
or random variations in the currents, will cause the transistors to go 
into one of the unstable states rather than the other. The circuit is 
therefore self-starting. 

The circuit subsequently oscillates between the two unstable states. 
The multivibrator is the only one of the three circuits which is a free- 
Page 255 


running oscillator, although if required pulses may be applied to 
synchronise the frequency. 

Positive feedback is provided by the capacitive cross-coupling from 
the collector of each transistor to the base of the other transistor. 
The loop gain is greater than one. Assuming an initial condition where 
Trl is on and Tr2 off (Fig. 1), then as C2i charges, the base b2 of Tr2 
goes negative and brings Tr2 into conduction. Regenerative switching 
occurs, and Trl is now off and Tr2 on. C2 2 now charges, bi goes 
negative until Trl conducts, and the cycle starts again. 

The duration t of the quasi-stable state — quasi-stable rather than 
unstable, since the circuit stays in this state for a certain length of time — 
is given approximately (Fig. 1) by 

t ~ 0-7C 2 R4 . 
This expression neglects, among other things, the fact that in reality 
C2 has two discharge paths, one through R4 and the other through the 
base of the transistor. When the transistor is cut off, its base current 
is constant and equal to the leakage current. The frequency of operation 
may therefore be somewhat temperature dependent. The output may 
be taken from either collector. 

The multivibrator is sometimes used in frequency multiplication, the 
required harmonic of the fundamental frequency being selectively 
amplified and appearing as a sinusoidal output from the tuned circuit. 

A representative multivibrator is given in Fig. 4, the values of CI, 
C2, R2 and R3 being chosen from the accompanying table accord- 





max. R2, R3 C1, C2 Typical p.r.f. 
(kO) (pF) (kc/s) 




Fig. A — Practical multivibrator 

ing to the transistor which is to be used. With these particular values, 
the circuit will operate with all transistors of the type in question at the 
stated pulse repetition frequency. 

Page 256 


The value of R2 and R3 is the maximum which ensures that the 
lowest-gain transistor will be bottomed when turned on. As the p.r.f. 
of the circuit is governed by the time-constant C1R2 ( = C2R3), the 
only possible modification is to increase the value of Ci and C2 if a 
lower operating frequency is required. 

In a symmetrical circuit, where C1R2 = C2R3 , the square waveform 
has equal on and off times. Unequal on and off times can be achieved 
by making the two time-constants unequal, but the circuit is not suitable 
for use with very large mark/space ratios. 

The limit to the p.r.f. is set by the switching times of the transistor, 
which are chiefly a function of cut-off frequency. It is only to be expected 
that, of the transistors considered in Fig. 4, the highest p.r.f. would be 
obtained with the OC41, but in some applications, the OC200 may 
have the advantage because of its very low leakage current. 

FLIP-FLOP (Monostable Circuit) 

The flip-flop circuit (Fig. 2) has one stable and one unstable or 
quasi-stable state. A trigger pulse flips the circuit into the unstable 
state, and the circuit subsequently Jlops back into the stable state. 

The astable circuit can be made monostable by replacing one of 
the capacitive cross-couplings by a resistive coupling. (In fact the 
coupling resistor is shunted by a capacitance, but this should be 
ignored for the moment.) The stable state is with Trl bottomed and 
Tr2 cut off. The trigger pulse switches Trl off and Tr2 on. The collector 
C2 of Tr2 goes positive, and takes the base bi of Trl positive with it, 
thus cutting off Trl. C2 now discharges through R4, and the circuit 
automatically switches back when the base-emitter voltage applied 
to Trl is approximately zero. 

The input capacitance of Tr2 reduces the loop gain at high fre- 
quencies, thus increasing the switching time. In practice the attenuation 
caused by the input capacitance is compensated by shunting R2 by CI. 

The duration of the quasi-stable state is again given approximately 
(Fig. 2) by 

t ~ 0-7C 2 R 4 . 

The (monostable) flip-flop delivers one output pulse for every input 
pulse. It may be used for pulse amplification and pulse shaping, or 
simply to delay the trigger pulse applied to the input, triggering of 
the following stages being effected by the rear edge of the output 
pulse. The output may be taken from either collector. 

Page 257 


A representative circuit for the (monostable) flip-flop is given in 
Fig. 5, with extra details in the table. The trigger pulse has to last 
long enough for the circuit to switch, that is, the trigger width has to 
be longer than the switching time. The length of the output pulse is 
determined mainly by the time-constant C3R3 • The value of C2 

(toon source) 

Trigger +1-5V 

Tr1, Tr2 

C2 t 
(pF) (us) 



OC71 4700 5 10,000 

OC200 1500 2 3300 

OC41 1000 1 1500 

Output-pulse length 


Fig. 5 — Practical flip-flop 

should be high enough to ensure that sufficient charge is extracted from 
the base of Trl when the transistor is switched off by the trigger pulse. 
Carrier storage is lower for transistors having higher cut-off frequencies, 
and a smaller trigger capacitance (C2) is therefore satisfactory for the 
OC200 and, to an even greater extent, the OC41. 


The original thermionic-valve version of the bistable circuit was 
described by Eccles and Jordan in 1919. There are two stable states, 
and the circuit will only change from one to the other when a trigger 
pulse is applied. 

The bistable circuit (Fig. 3) can be derived from the astable circuit 
by replacing both the capacitive cross-couplings by resistive couplings. 
The two resistors are in fact shunted by capacitors, for the same 
reason as in the monostable circuit. 

The two stable states of the circuit are : 

(a) Trl bottomed and Tr2 cut off; 

(b) Trl cut off and Tr2 bottomed. 

If Trl is bottomed and Tr2 cut off, a positive trigger pulse on the base 

Page 258 


of Trl will tend to cut this transistor off. The collector ci of Trl will 
go negative, and the base D2 of Tr2 will go negative in turn. Tr2 there- 
fore comes into conduction, and C2 and hence bi go more positive. 
Thus the loop can be made regenerative, the necessary condition being 
that the voltage gain from base to base be greater than one. 

The choice of the various component values is always a compromise. 
A conventional bistable circuit with representative values is shown in 

convert basic 
bistable circuit 
to a binary 

Tr1 , Tr2 t C3, C4 Typical trigger p.r.f. 

~™ fe?> <P F > < kc ' s > 

OC71 10 4700 20 

OC200 2 1000 80 

OC41 1 470 200 

Fig. 6 — Practical bistable circuit 

the top part of Fig. 6; however, for some particular application, it may 
be necessary to change the values considerably. 

Each trigger pulse causes the circuit to change from one stable state 
to the other. The bistable circuit carries out one complete cycle for 
every two input pulses. During a complete cycle, one pulse appears 
at each collector. The pulse repetition frequency of the output from 
either collector is therefore half that of the input. (This is the basis of 
binary counting.) 

The simplest method of taking off the output is by capacitive 
coupling from either of the collectors. A higher switching speed may 

Page 259 


be obtained, however, with a coupling transformer connected in series 
with one of the collector load resistors. With transformer coupling, 
repetition rates of up to approximately one-fifth of f a are possible. 

One sound method of converting the bistable circuit to a binary 
divider (binary counter) is illustrated by the components boxed in by 
the dashed line in the lower part of Fig. 6. This circuit has been 
designed for reliable operation with all transistors of the type in 
question up to the stated values of trigger-pulse repetition frequency, 
so that these values do not represent the ultimate in performance. 

The diodes Dl and D2 are controlled by the collector voltages. 
When Trl is on, it is bottomed, so that Dl is just conducting. Tr2 
is off, and D2 is biased off by the negative voltage at the collector. 
The next positive pulse is passed to the base of Trl, switching Trl off 
and Tr2 on. The time-constant C3R7 ( = C4R8) is arranged such that 
the diode remains conducting until the end of the input pulse, thus 
ensuring that the circuit will always complete its switching before the 
'gate' is opened to the next input pulse. 

The time that must be allowed to elapse between trigger pulses is 
determined by the time required for C3 (or C4) to return to its initial 
voltage. Thus the input time-constant C3R7 = C4R8 sets an upper 
limit to the trigger-pulse repetition frequency. For effective triggering, 
this time-constant needs to be about five times the trigger width. 

The typical p.r.f. is of course highest for the OC41 which, with its 
higher-frequency version the OC42, is recommended for this type of 

The pulse repetition frequency may be divided by any power of two, 
by combining the requisite number of binary counters in cascade 
(binary system). 


The Eccles- Jordan bistable circuit is completely symmetrical. For some 
purposes an asymmetrical circuit may be preferable. Such a circuit is 
shown in Fig. 7. 

Trl is a grounded-base and Tr2 a grounded-collector stage. In the 
grounded-base configuration, the current amplification factor is 
almost one, the input impedance is low and the output impedance 
is high. At low frequencies, the transistor does not introduce any phase 
shift between the input and output signals. To obtain regenerative 
feedback from the output to the input, it is necessary to bring about 
a transformation in impedance without introducing any change in 

In the grounded-collector arrangement, a transistor has a high 
input and a low output impedance. It has a current amplification 

Page 260 


factor of a' and at low frequencies there is again no phase shift within 
the transistor. 

A combination of two such stages therefore gives regenerative feed- 
back. The loop gain is greater than unity, since R 2 > R4 . 

The two stable states of the circuit are: (a) Trl conducting, Tr2 
almost cut-off; (b) Trl cut-off by a negative bias on its emitter, Tr2 

O IjLisec 4>tsec 

Fig. 7 — Asymmetrical bistable circuit 

conducting. The circuit may be triggered on either the base or emitter 
of Trl, both points being sensitive to pulses of either polarity. 

The two diodes Dl and D2 are added to the circuit to prevent 
excessive variation in trigger sensitivity from circuit to circuit. They 
also improve the switching time. Dl is a catching diode which prevents 
Trl from bottoming, thus greatly reducing the hole storage. A smaller 
trigger pulse is required, since it is no longer necessary to expel so 
many holes from the base, and at the same time the switch-off time is 
shortened. D2 limits the cut-off bias and thus reduces the negative 
trigger pulse required on the base to switch on the transistor. The 
2mH choke is a further improvement which increases the high-frequency 
loop gain and improves the switching time by about lfxsec. 

The asymmetrical bistable circuit is suitable for a repetition rate 
of about one-fifth of f a . 

Page 261 



The OC139 and OC140 are n-p-n transistors and are approximately 
opposite-polarity versions of the OC41 and OC42 respectively. The 
output from the collector of a p-n-p transistor is of suitable polarity 
for direct coupling to the base of an n-p-n transistor. Fig. 8 shows 

Fig. 8 — Bistable circuit using p-n-p and n-p-n combination 

such a bistable circuit using the OC41 and OC139. The current gain 
is greater than one and there is no phase reversal. 

The base of the n-p-n transistor, Tr2, is connected by a series resistor 
to a voltage more negative than the collector supply voltage. This 
arrangement allows the base current to reverse, and improves the 
switch-off time. 

Dl is a catching diode which prevents bottoming of Trl. 


The asymmetrical bistable circuit (Fig. 7) may be used as a Schmitt 
trigger circuit. The input is applied to the base of Trl via the existing 
IkQ, resistor Rl. The circuit will then trigger in one direction when the 
input exceeds a certain level, and will trigger back again when the input 
returns to approximately the same level. 

The backlash or hysteresis of the system depends on the loop gain 
and only disappears when the loop gain is just equal to one. 

The p-n-p/n-p-n bistable circuit (Fig. 8) may be used in a similar 
manner, the input being applied to the emitter of Trl. 


Bistable, monostable and astable circuits may be designed using 
transformer coupling to provide the positive feedback, instead of RC 

Page 262 


coupling. The transformer increases the loop gain, and hence results 
in faster switching. 

Blocking oscillators as such are transformer-coupled monostable 
or astable pulse generators. Whether or not the circuit is free-running 
(astable) depends on the d.c. conditions. If a d.c. state exists in which the 
loop gain of the circuit is less than one, then the circuit will be stable 
in that state and will require to be triggered out of it (monostable 

The circuit consists essentially of a transistor with transformer- 
coupled feedback from the collector to either the base or the emitter, 
and with an RC timing network in either the emitter or the base 
circuit. Sufficient feedback is applied to drive the transistor into bottom- 
ing at the beginning of the first cycle of oscillation. The transistor then 
remains bottomed for a period determined by the circuit constants, 
after which regeneration causes a rapid switch-off. The output voltage 
takes the form of a rectangular pulse. 

The advantages of the blocking oscillator are a low output impedance, 
short output pulses if required, and an output which can be adjusted 
to any required amplitude by a suitable choice of transformer winding. 

Triggered Circuit 

A practical circuit for a triggered blocking oscillator is given in Fig. 9. 
This circuit will operate at a pulse repetition frequency of lOOkc/s at 









Fig. 9— Triggered blocking oscillator 

ambient temperatures of up to 60°C. 

The transformer details are: Ferroxcube core, FX1011; primary 
(collector) winding, 60 turns of 38 s.w.g. wire; secondary (emitter), 
winding, 10 turns of same wire. 

A diode and a damping resistor R4 are connected across the primary 
of the transformer to prevent excessive overshoot. Bias is applied by 
Rl and R2, CI is the coupling capacitor, and R3 ensures a certain 
minimum source impedance for the transistor. 

Page 263 


During the bottomed period, when the collector voltage is constant, 
the collector current continues to increase at a constant rate through 
the inductive branch of the load. The emitter current however is vir- 
tually constant, and is determined by the transformer feedback voltage 
and the resistance in the emitter circuit. When the emitter current is 
no longer sufficient to maintain the rate of increase of collector current, 
the voltage across the transformer falls; the transistor comes out of 
bottoming, and regeneration causes the voltages across the transformer 
windings to reverse and the transistor to switch off rapidly. 

C2 then discharges through R2, and the energy in the transformer 
is dissipated in the diode and damping resistor. The larger of these 
time-constants determines the period before the cycle can recommence. 

Free-running Circuit 

Fig. 10 shows a free-running blocking oscillator which can operate 
as a frequency divider. If the oscillator is synchronised by an input 
p.r.f. of up to lOOkc/s, a p.r.f. of lOkc/s, or proportionately less, may 

Fig. 10 — Free-running blocking oscillator (frequency divider) 

be obtained at the output. The division ratio may be as high as 10:1 
and is stable for ambient temperatures of up to 60°C. The transformer 
core is again the FX1011 and the primary (collector) winding 60 turns 
of 38 s.w.g. wire, but the secondary (base) winding is now increased to 
30 turns of the same wire. 

Despite first appearances, this circuit is essentially a variation on 
the preceding one. The feedback is applied to the base instead of to 
the emitter, as the loop gain is then higher and the circuit oscillates 

Page 264 


more freely. The timing circuit is in the emitter circuit, RV3 being the 
frequency control. 

The prime requirement of a frequency divider is that the reset time 
be stable. The stability of this time determines the maximum division- 
ratio of the circuit. 

The reset time of the circuit is the time required for the capacitor 
to return to a voltage at which the transistor can conduct again. In 
addition to the discharge through the associated resistance, the leakage 
current I e0 (o) helps to discharge the capacitor. As the leakage current 
is temperature dependent, for a stable reset time it is essential to ensure 
that Ieo(o) at its maximum value is a negligible part of the total dis- 
charge current. 

If the RC timing network is in the base circuit, the leakage current 
which contributes to the discharge is the collector leakage currently (o>- 
With the timing network in the emitter circuit, only I e0 (o) flows to 
the capacitor. As I e o(o) is only about one-fortieth of I CO (o), the tempera- 
ture dependence of the circuit is very much reduced. 


Transistors may be used to generate sawtooth-waveforms in circuits 
which take their names, at least, from thermionic-valve counterparts. 

A well-known technique in thermionic-valve circuits is the use of 
an anode-grid capacitor to provide Miller voltage feedback. A feed- 






1 r 



WW o 

15kf> " 3V 

Fig. 11 — Self-gating Miller circuit 

back capacitor may likewise be used with transistors to provide current 

The discharge of the capacitor can be made sufficiently linear to 

Page 265 


provide a sweep voltage for a cathode-ray tube. For a fast flyback, it 
is necessary to interrupt the negative-feedback action of the circuit. 
One method is to connect two transistors in series, the feedback loop 
being open or closed according to whether the second transistor is 
cut-off or conducting. A further transistor, used instead of a catching 
diode, makes the circuit self-gating. 

A practical self-gating Miller circuit is given in Fig. 11. Trl is the 
catching transistor; Tr2 and Tr3, which are connected in series by R3, 
provide two gating electrodes in the current path and simulate, to 
some extent, the gating action of a pentode. 

In the rest condition, Trl and Tr3 are conducting, and the collector 
of Tr3 is at approximately — 1-5V, the base potential of Trl. Both the 
diodes Dl and D2 are conducting, so that the base and emitter of Tr2 
are both at the same potential (— 3V) and the transistor is just cut off. 
The collector of Tr2 is therefore at the line potential of —18V. 

On applying a trigger pulse to cut off Trl (for example, a negative 
trigger pulse on the emitter), V c i switches to — 3V. Dl and D2 are both 
cut off; Tr2 conducts, feeding back a step of current to the base of 
Tr3 via CI. The capacitive feedback loop through Tr3, Tr2 and CI is 
thus closed, and the current through these two transistors rises linearly 
with time. The positive-going flanks of the sawtooth occur at the collect- 
ors of Tr3 and Tr2. When V c3 reaches — 1-5V, Trl starts to conduct; 
V c i goes positive, cutting off Tr2, so that V C 2 flies back to —18V. The 
circuit thus resets itself to the rest condition. 

In addition to the sawtooth waveform at the collector of Tr2, a 
rectangular waveform, corresponding to the sweep time, is available 
at the emitter. The circuit may be triggered by a positive pulse applied 
to the collector of Tr2, or by a negative pulse at either the collector or 
emitter of Trl. 

The circuit illustrated is only one among several possible circuit 
combinations. A parallel instead of a series arrangement may be 
adopted for the gating transistor, and a directly coupled p-n-p/n-p-n 
combination (giving a faster flyback time) is suitable for the feedback 
loop. For a particularly linear sweep, a circuit may be developed from 
the thermionic-valve 'bootstrap' circuit. The need for suitable trigger 
points is another factor which influences the choice of circuit arrange- 

Page 266 



Amplification of a d.c. input is frequently required for industrial and 
scientific purposes. Many applications will probably suggest them- 
selves to the worker in any particular subject: 

(a) d.c. amplifiers may form part of temperature-measuring devices 
using thermocouples, platinum resistance thermometers or 

(b) in general, the d.c. amplifier permits low-range meters, such as 
micro-ammeters and high-impedance voltmeters, to be replaced 
by more robust types 

(c) once the d.c. level has been raised sufficiently, it can be used to 
operate a relay in an alarm or automatic-control system 

(d) other applications include analogue computers and electro- 

Amplification of a d.c. signal is a simple requirement, but the choice 
of a suitable circuit can be a difficult problem, as many of the circuits 
which have been published deal with specific applications. This chapter 
gives a survey which will make it fairly simple to choose the type of 
d.c. amplifier required. 

A comparison of the performance of the circuits is given at the end 
of the chapter. The circuits illustrate the principles of the various types 
of d.c. amplifier, and usually modifications will be necessary for any 
particular application. 

Directly coupled and chopper types are discussed. 


The main problems of d.c. amplification are drift of the zero-reading 
and variations in gain. The gain can be kept as constant as required 
by applying negative feedback. Feedback does not dispose of drift, 
however, as the gain and drift are reduced proportionally. 

Temperature variations are mainly responsible for drift in transistor 
circuits, and their effect is particularly pronounced in directly coupled 
amplifiers. In these amplifiers, the temperature dependence of the 
leakage current and of the base-emitter voltage is most important. 
The temperature dependence of a' is not usually significant, though in 

Page 267 


some circuits it has to be taken into account. These effects are additive, 
that is, they all cause an increase in collector current with temperature. 

The effect of the base-emitter voltage Vbe is less well known than 
that of the leakage current. Vbe at constant emitter current falls by 
roughly 2mV per °C rise in temperature, for both silicon and germanium 
transistors made by the alloy-junction method. If the stage is fed from 
a low source impedance, the drift introduced by the change in Vbe as 
the temperature varies may be larger than that introduced by the change 
in leakage current. In the simpler type of directly coupled amplifier, the 
minimum source resistance is specified at which the drift caused by 
Vbe becomes comparable with leakage-current drift. In more elaborate 
arrangements, this drift is taken care of in other ways. 

Amplifier using Germanium Transistors 

Fig. 1 shows a straightforward experimental circuit using two german- 
ium transistors (an OC72 and an OC35). Grounded-emitter connection 
is the obvious choice for direct coupling. The amplifier should be 

Fig. 1- 

-Directly coupled amplifier using germanium transistors. Supply —12V 

tapped at — 8V. Current drain 1-3A for full-scale output. OC35 passes 

constant 0-3A. 

driven from a reasonably high source impedance ( > 3kO) to reduce 
the Vbe drift to the same level as leakage-current drift. A set-zero 
potentiometer is included, and a tap on the supply facilitates a zero- 
current reading. 

Amplifiers of this type might be used as the output stages following 
a more sensitive amplifier after the signal level has been raised suffi- 
ciently compared with the maximum drift. 

Amplifier using Silicon Transistors 

The leakage current is much lower for silicon than for germanium 

Page 268 


Drift is therefore very low for the amplifier using silicon transistors 
(Fig. 2), and as the leakage current no longer masks the temperature 
dependence of Vbe and a', these become of greater significance. 

A source resistance of at least 300kQ should be provided for the 
amplifier for the Vbe drift to be negligible. It is likewise necessary to 
provide a high source resistance for the second stage. To use the 
requisite high load resistance ( = 33kO) in the first stage, a supply 
voltage of at least 24V is required. 

The change of a' with temperature is no longer negligible. Negative 
feedback is incorporated which reduces the drift caused by variations 
in a' (and also the gain) by a factor of about 8. The feedback is 
provided over three stages, not merely the first stage, and is thus more 
effective because of the higher gain over the three stages. The feedback 
also stabilises the operating points of all three stages, not the first 

o — 1 ^g) 





Calibrate ^ Zero 

- J_ OAZ204 

. \ - 1" 4-7V . [ - ~X" 6-8v 


Fig. 2 — Directly coupled amplifier using silicon transistors. Supply —24V at 
60mA. Zener diodes equivalent to voltages shown. 

stage alone. The feedback resistance should not be made so low that 
changes in Vbe begin to introduce appreciable drift. 

Direct coupling brings a further difficulty. The feedback resistance 
must establish the operating point and also provide the feedback. In 
this circuit, the resistance required for bias was considerably higher 
than that required for feedback. An emitter resistor is therefore included 
in the first stage. 

The resistance in the emitter circuit raises the base potential to a 
few volts, so that a reverse current flows through the source. In some 
circuits this may not be permissible. Another arrangement would be to 
connect the source between base and emitter. 

Page 269 


The effective gain of the amplifier is 1000 with feedback, but depends 
on the load resistance, which can be varied to adjust the gain. 

Operating the transistor at very low voltages, comparable with Vbe, 
would make the gain sensitive to changes in the Vbe of the next stage. 
Hence Zener diodes are used in the emitters of the second and third 
stages to ensure that sufficient voltage is applied to the collectors of 
the first two stages, so avoiding the loss of gain which would occur if 
emitter resistors were used. 

Long-tailed-pair Amplifier using Germanium Transistors 

Both the amplifiers described so far require a high or relatively high 
source resistance. For amplifying signals from low-resistance sources, 
some method must be found of eliminating drift introduced by Vbe 
( ~ — 2mV/°C). Because this drift is very nearly identical for any two 

kOS kfi; 


" J 'T" "^ 

1-2> ! 


t i- ^ 

I OC7] to 5V 2i _^-" 

f<) Kv>^vC^ 

V_ST j Calibrate , 

Fig. 3 — Long-tailed-pair amplifier using germanium transistors. Supply —12V 

at 20mA. 

transistors of any given type (say, two OC71), it can be cancelled by 
using another transistor of the same type to balance out the change. 

The resulting circuit is the long-tailed-pair amplifier, an example of 
which is shown in Fig. 3. In this circuit, the error arising from differences 
in leakage current can be made small by using a low source resistance 
(< 100Q). The set-zero resistance in the emitter circuit introduces 
some unbalanced feedback and is therefore kept as low as possible. 
For reasonable stability, the voltage drop across the shared emitter 
resistors has to be much greater than the change in Vbe- 

The long-tailed-pair circuit is effectively a method of applying 
differential negative feedback so that in-phase changes produce no 
output, an output only occurring with push-pull signals. 

Page 270 


Long-tailed-pair Amplifier using Silicon Transistors 

The leakage current of silicon transistors is very small, and does not 
merit consideration over the usual ambient-temperature range up to 
35°C (95°F). Differences in Vbe and a' can introduce drift, however. 

Any source resistance can be used, since it is not limited, as when 
using germanium transistors, by the need to eliminate differences in 
leakage current. The biasing arrangement is therefore changed to that 
shown in Fig. 4. The d.c. conditions in the first two transistors are 

Fig. A — Long-tailed-pair amplifier using silicon transistors. Supply 


-12V at 

set by two potential dividers, which do not significantly shunt the input 
resistances, although they have the usual stabilising effect on the 
operating point. The zero under open-circuit conditions is adjusted by 
varying one of the bias resistors. The zero adjustment under short- 
circuit conditions is made as before by setting the potentiometer in 
the emitter circuit. 

The drift with open-circuit input for this amplifier is the same as for 
the one using germanium transistors with a restricted source resistance. 

Temperature Control 

Changes of temperature are mainly responsible for drift. Drift can be 
very much reduced by using a circuit such as that shown in Fig. 5 to 
maintain the temperature of the amplifier at a constant level. The 
amplifier is mounted in a metal block drilled to take the transistors. 
A closed-loop feedback system controls a power transistor, which 
supplies the current for heating a coil wound uniformly around the 
metal block. The sensing element used to provide the error signal is 

Page 271 


an a.f. transistor (the first OC71) mounted in the block. The leakage 
current r co of this transistor changes markedly with temperature, and 
the changes, after further amplification, control the current through 
the heating coil. 


For applications where drift requirements are more stringent, direct 
voltages or currents can be converted to proportional a.c. signals : this 
technique is known as chopping. The first amplifiers built on this 

Trl ,^^IOC71 


Tr3 >— JOC72 



— yyw> — " 


Fig. 5 — Simple temperature-control circuit. Supply —12V at 0-3mA. 

principle made use of mechanical switches in which a vibrating reed 
interrupted the d.c. signal periodically. 

The problems associated with chopper-type d.c. amplifiers are 
centred on the design of the actual chopper, differing slightly for 
mechanical and semiconductor choppers. 

Mechanical Choppers 

The design of mechanical choppers will not be described here. The 
maximum practical speed of operation is at present about 400 to 

Semiconductor Choppers 

The speed of operation of semiconductor choppers is appreciably 
higher than for mechanical types, the limit being set by minority- 
carrier storage. Chopping frequencies of about 1000 to 1500c/s are 
commonly used at present. The higher speed makes possible a wider 
amplifier bandwidth. 

Semiconductor choppers are still somewhat inferior to the best 
mechanical choppers as regards zero drift. The advantages of rugged- 

Page 272 





" — yv\AA- 











Page 273 




^■i oooooi rooooo'i rwowi 


- -O 

< r 







— A/V^A — y^VV — K — ''05550"'— 



I 2§ 

1 si 












































LO o 


+ 1 


























































Page 274 


ness and long life, however, make the former superior for continuous 
operation and for mobile equipment. 

Silicon-diode Chopper 

The amplifier shown in Fig. 6 incorporates a bridge chopper which 
uses silicon diodes. OA202 silicon diodes are completely satisfactory 
in this design. 

LI and L4 are 130 turns, and L2 and L3 are 30 turns, all of 35 s.w.g. 
Lewmex wire. L5 is 400 turns of 38 s.w.g. Lewmex wire. L6, the 
detector choke, has an inductance of 3H and a d.c. resistance of 90&. 
All formers are Ekco type DP10857, and the laminations are Mu-metal 

Single-transistor Chopper 

Fig. 7 shows an amplifier using a transistor chopper. The stability 
is significantly superior to that of other purely semiconductor amplifiers 
and approaches the performance of amplifiers using mechanical 
choppers. The amplifier was designed to amplify very low currents 
from high-impedance sources. 

In its present form the amplifier is rather susceptible to changes in 
supply voltages to the driving-waveform circuits. Use of a slightly 
lower chopping frequency and omission of the delay circuit can over- 
come this drawback. The d.c. amplifier as a whole would then have 
reduced bandwidth, which is not always a disadvantage. It should be 
possible to make a significant improvement to this type of amplifier 
using silicon alloy-junction transistors such as the OC200 and OC201. 

With OC71 transistors in the a.c. amplifier, the output is not identical 
for identical inputs of opposite polarities. This drawback can be 
remedied where necessary by using OC45 transistors in the first three 

Balanced Transistor Chopper 

In the circuit shown in Fig. 8, two transistors (Trl, Tr2) are used as 
a balanced chopper. The input is applied to only one of the transistors, 
while the error voltages of the two transistors appear in opposition.' 
Balance is effected by setting the lkD. potentiometer RV5 to give a 
zero output reading for zero input. 

The complete circuit shown in Fig. 8 is intended for amplifying 
voltages of less than ±200(xV from low-resistance sources (for example, 
thermocouples and strain gauges). 

The a.c. amplifier consists of Tr3 to Tr6. A synchronous detector 
Tr7 gives an output in the same phase as the input. The chopper and 

Page 275 


'~t s 


Page 276 


the detector are driven by a multivibrator (Tr8, Tr9) operating at a 
frequency of 1650c/s. 

If the output signal is required as a current, the amplifier consists 
of three OC45 and one OC72 (Tr6), as shown in Fig. 8. The detector 
is then loaded by an ammeter which gives full-scale deflection at 0-5mA. 
With this arrangement the meter reads about one micro-amp per 
microvolt. If a voltage output is preferable, the a.c. amplifier consists 

5 r 




R4'0 C22 

-A/WV— 1| — 

180 n. o-l^iF 

-o Output 


-° JTTL 


Fig. 9 — Alternative final amplifier stage (Tr6) for voltage output from balanced- 
chopper amplifier. 

of four OC45, and the final stage is arranged as shown in Fig. 9. This 
circuit allows an output of 750mV to be obtained for an input of 400(jlV. 
This voltage may be fed into the input of a valve amplifier, or into an 
oscilloscope to display slowly varying low voltages. 

Mechanical-chopper Amplifier 

Fig. 10 is the circuit diagram of a chopper-type amplifier using a 
mechanical chopper which short-circuits the input periodically. The 
chopper is a Carpenter 3PK55 polarised relay supplied specially 
adapted for use as a chopper. It is driven by a simple multivibrator 
at a frequency of about 400c/s. The resulting square wave is amplified 
in a transformer-coupled grounded-base amplifier, to which two 
RC-coupled grounded-emitter stages are added to raise the signal to 
a suitable level for rectification and for driving a meter. 

Very careful attention must be paid to the screening of the chopper 
and of the input to the amplifier. 


The performance of the amplifiers described in this chapter is com- 
pared in Tables 1 and 2. No claim is made that the amplifiers give the 

Page 277 






a. <u 
a. a. 
=j a. 
(/) o 


Page 278 


best possible performance, and therefore component tolerances have 
not been considered. The figures quoted are typical of what can be 
achieved using average transistors in the circuits described. 


Drift for the various semiconductor types is quoted in terms of changes 
in temperature. For the mechanical chopper, drift is essentially a 
question of mechanical design, and is therefore quoted in /uV per hour. 

Where chopper types are concerned, the mechanical chopper offers 
the best performance in terms of zero stability. 

The source resistance of a directly coupled amplifier either has to 
be very low or very high if minimum drift is to be achieved. The choice 
of d.c. amplifier will be governed to some extent by the source resistance 
it is proposed to use. For example, for thermocouples, which typically 
have a source resistance of less than 100Q, a long-tailed-pair amplifier 
is the obvious choice. 

Silicon transistors offer a great improvement for current-fed ampli- 
fiers, where the source resistance is much greater than the transistor 
input resistance, as the figures in the first two columns of Table 1 show. 
Silicon types do not offer much advantage where a low-resistance source 
is specified, as can be seen by comparing the figures in the third and 
fourth columns for the long-tailed-pair circuit. 

Drift for a long-tailed-pair amplifier can be reduced by a factor of 
about 20 by stabilising the temperature. A similar improvement would 
be obtained if any of the other directly coupled amplifiers were mounted 
in the temperature-controlled block. The supply must be correspond- 
ingly stabilised to obtain the full advantage of temperature control. 
Drift in a directly coupled amplifier with temperature control is com- 
parable with that of chopper types. 

The circuitry becomes correspondingly more complicated as the 
drift requirements are made more stringent, consequently the simplest 
circuit should be chosen which is adequate for the application in mind. 
If milliamps are to be amplified, it is not necessary to design a circuit 
with a drift of the order of millimicroamps. 

Page 279 


Performance of Directly Coupled Amplifiers 



Long-tailed Pair 

Type of Transistor 

Germanium Silicon 











Zero drift 

4 to 8 










Full-scale output 






Specified source 


>300kft < 100ft 


< 100ft 

Input resistance 






Load resistance 






Performance of Chopper-type Amplifiers 

Type of Chopper 










Zero drift 

m(iA/ C 




2 to 3 

Full-scale output 





Specified source resistance 





Input resistance 





Load resistance 





*Measured for a change in temperature from 20 to 35°C 
Page 280 



In many of the d.c./d.c. converters which have so far been published, the 
switching is controlled by increasing the collector current of the 
transistors. Most of the circuits described in this chapter belong to a 
family of possible arrangements for controlling the instant of switching 
by decreasing the base current. All these circuits use two transformers. 

Details are given of the design and performance of a high-power 
push-pull converter capable of 100W output. 


Transistor d.c./d.c. converters compare very favourably with rotary 
converters, vibrator-transformer-rectifier converters and mechanical 
methods generally for converting energy from one level to another. 
Without exception such methods are inefficient at low powers. At 
higher powers, where their efficiency is fairly good, the mechanical 
systems still suffer from a number of disadvantages, namely: high 
initial cost; bulk and weight; maintenance costs; and interference 
from arcing at the contacts. (Nevertheless, the transistor converter 
needs to be adequately screened.) 

Transistor d.c./d.c. converters perform the same sequence of opera- 
tions as the vibrator-transformer-rectifier circuit. First, the d.c. from 
the battery is chopped, using a square-wave oscillator. The chopped 
input is then stepped up to a higher level. Finally, the stepped-up 
signal is rectified to convert it back to d.c, and the output is smoothed 
and delivered to the load. 

Usually some bias is provided, at least when first switching on, to 
start the oscillator. Further components may be incorporated to 
improve the regulation, and conventional voltage multipliers are used 
to step up the voltage on the output side. 

Long leads to the supply possess appreciable inductance, and the 
supply will have to be smoothed by a capacitor sited near to the con- 


The two-transformer converter (as in Fig. 1) overcomes various 
disadvantages of the usual one-transformer type. The new circuit differs 

Page 281 


from the conventional type of push-pull converter, in that a small 
saturating drive transformer is used to control the switching, and a 
larger transformer, working linearly, steps up the output to the required 
value. The essential improvement is that a higher proportion of the 
transistor peak-current rating can be used with all transistors, and 
therefore the output power can be increased, while tolerating the full 
production spreads in the characteristics of the transistors. 

On connecting the supplies to the circuit in Fig. 1, one of the transis- 
tors (say Trl) will conduct, because of the unbalance in the circuit, 
causing its collector voltage to swing (to zero) by very nearly the 

Fig. 1 — Practical two-transformer converter 

supply voltage. The voltage building up across the primary of the 
output transformer is applied across the primary of the drive trans- 
former Tl in series with a feedback resistor Rf . The secondary windings 
are so arranged that Tr2 will be reverse biased and will remain cut off 
and Trl will be held in the bottomed condition. 

As soon as the core of Tl reaches saturation, rapidly increasing 
primary current causes an additional voltage drop across the feedback 
resistor Rf. This drop reduces the drive; and the collector current of 
Trl, which was bottomed, starts to decrease, causing the polarities of 
the voltages in all the windings to reverse. Trl is rapidly driven to cut 
off and Tr2 switched on. Tr2 continues in this state until the negative 
saturation of the transformer is reached. 

The circuit switches back to the initial state and the cycle is repeated. 
The oscillation then continues at a frequency determined by the 

Page 282 


design of the saturating transformer Tl and by the value of the feedback 

For reliable starting, the transistors are initially biased into conduc- 
tion by using a resistor and a diode (Rl and Dl, Fig. 1). The external 
base resistors are added to reduce the effect of Vbe on the operation of 
the circuit. 

The collector current in either of the transistors rises to the load 
current, plus the magnetising current of the output transformer, plus 
the feedback current needed to produce the drive. Because the output 

I c max.— 
I c 

Light load 

Fig. 2 — Collector voltage and current waveforms for purely resistive load 

transformer is not allowed to saturate, the magnetising current is only 
a small fraction of the load current. 

The collector voltage and current waveforms, for a purely resistive 
load, are shown in Fig. 2. 


The design of a converter is normally based on the available supply 
voltage, the required output voltage, and the output power. 

The peak voltage at the collector of either transistor, when cut off, 
is approximately twice the supply voltage. The supply voltage should 
therefore always be less than half the collector breakdown voltage at 
the peak value of inductive current. 

The design of the transformers is not critical, and a wide choice of 
operating frequency can be tolerated, depending on the required size 
and weight of the converter and its efficiency. 

Drive Transformer 

The primary of the drive transformer, in series with a feedback resistor 
Rf , is connected across the two collectors of the transistors of the 
converter (Fig. 1). 

The peak voltage produced across the two collectors by the primaries 

Page 283 


of the output transformer is approximately twice the supply voltage. 
The voltage applied across the primary of the input transformer, 
however, depends on the value of the feedback resistor and the required 
drive current. The value of the feedback resistor, in turn, is a comprom- 
ise between the requirements of the saturation current of the trans- 
former, the voltage applied across the primary, and the operating 
frequency of the converter. 

The number of turns N p required for the primary winding is deter- 
mined as follows. The expression relating the operating frequency (f) 
and the various transformer and circuit parameters is : 

f=^I2! ... a) 

4NpAB s l ' 

Vm is the voltage applied across the primary 
N p is the number of turns on the primary 
B s is the flux density at saturation in gauss 
A is the cross-sectional area in square centimetres. 

Another condition for correct operation is that there should be sufficient 
current available to saturate the core. This condition is given by a 
commonly used transformer equation: 

„ 4tcNI 1-26NI 

H = ~ • 

10/ I ' 

from which, for saturation: 

Hs= 1^6NplL >Hoj (2) 


H s is the strength of the magnetising field at saturation, expressed 
in oersteds 

H is the intrinsic strength of the magnetising field of the material 
used for the core, expressed in oersteds 

II is the inductive current in amperes 

/ is the length of the magnetic path in centimetres. 

In Eq. 2, H s is fixed by the material chosen for the core, and / by the 
size of the laminations. Therefore a value for N p can be obtained for a 
particular inductive current II . 

This value of N p can now be substituted in Eq. 1 to obtain the 
operating frequency. If a different frequency is required, adjustment 

Page 284 


of the cross-sectional area is necessary, which entails a change in the 
number of laminations. 

The number of primary turns having been decided, the feedback 
winding Nf is designed from the expression: 

N f = N p .^? ...(3) 

where Vf , the required feedback voltage, is given by 

Vf = Vbe+lBRb+Vm , . . .(4) 


Vbe is the base-emitter voltage for peak collector current 

Ib is the base current required for peak collector current 

Rb is the external base resistance 

Vdi is the forward voltage drop across the starting diode Dl. 

Output Transformer 

The output transformer, T2, is a normal linear transformer and is 
designed using conventional techniques. The primary windings must 
have a sufficiently high inductance to keep the required value of 
magnetising current low. Also, the leakage inductance must be made 
negligibly small, by using bifilar windings. 

The inductance required for each half of the primary can be 
calculated from 

L = V cc .ii, ...(5) 

dl m 

Vcc is the supply voltage 
t is the time of half a cycle 
i m is the magnetising current. 

Starting Circuit 

The basic circuit arrangement will not necessarily start to oscillate, 
especially when heavily loaded, because both transistors are initially 
cut off. A permanent bias is therefore applied, by means of Rl and Dl, 
so that the circuit has a loop gain greater than unity and will always 
start to oscillate. 

To ensure a loop gain greater than unity, the base current lb (in mA) 
must be greater than 

2^ , ...(6) 

(a'R L '-nR bb ) 

Page 285 



R L ' is the resistive load appearing across the primary winding 
Rbb is the total base resistance, both internal and external 
n is the turns ratio of the feedback winding. 

The value of Rl can be calculated approximately from 

Rl= Vcc , ..-(7) 

2I b +Id 

where Id is the inverse current of the diode. For silicon diodes Id is 
usually small enough to be neglected. 

It is possible to use a resistor instead of the starting diode but, if 
the value of the two starting resistors is high, the drive power will 
need to be increased substantially. Small values for the resistors 
increase the current drain and so lower the efficiency. 

If a resistor, R2, is used instead of the diode Dl, the value of Rl can 
be found from 

Vnc ...(8) 

Rl ~ 

~ T . Vbe + IbRbb 
21b -f 


where Vbe is the base-emitter voltage for the required base current, 


Several factors affect the design of a practical two-transformer con- 
verter; these factors and the performance of the circuit are now 
examined in detail. 

Operating Frequency 

The choice of operating frequency is not very critical and will depend 
on the efficiency and physical size of the converter. 

Although losses in the transformer cores and transient losses of the 
transistors increase with operating frequency, the efficiency varies 
only a few per cent over the frequency range 300 to lOOOc/s. 

Feedback Resistor 

The optimum value of the feedback resistor Rf is found, experimentally, 
to be that value which will drop about half the available voltage, at 
the drive current corresponding to the maximum load current. 

Increasing Rf causes a greater drop in voltage across it, so that less 

Page 286 


voltage is available across the primary of the drive transformer. As 
inferred from Eq. 1, the operating frequency will decrease. 

Decreasing Rf will increase the operating frequency and increase the 
losses arising (a) from the circuit resistance and (b) in the transformer 
core, because of the higher magnetising current. 

Drive Transformer 

Since the required drive is less than one watt, only a small core is 
needed. A square stack of Telcon HCR alloy laminations (pattern 224) 
can be used. This material has the following characteristics: 

B s = 15,000 gauss, H s = 2 oersteds, / = 5-72 cm. 

The cross-sectional area (A) of 50 laminations equals 0-331 square 

If the magnetising current II is 40mA then, from Eq. 2, 
_ 2x5-72 _ 
"*~ 1-26x40x10-3 _22/ - 

Thus there should be 227 turns on the primary winding. 

To evaluate Eq. 1, the values of the primary voltage Vm and of the 
feedback voltage Vf must first be calculated. 

The maximum base-emitter voltage Vbe max required for lower-limit 
OC28 transistors is 1-6V, and the maximum base current Ib max is 
375mA for the maximum peak collector current of 6A. With an external 
base resistor of 10Q, and allowing IV across the starting diode, the 
feedback voltage required at 375mA is 6 -35V (from Eq. 4). 

If a turns ratio of 4 : 1 is chosen, 57 turns are required for each 
feedback winding, and the primary current I p is about 94mA. The 
voltage developed across the primary under these conditions is 

Vi„ = n.Vf 
= 4x6-35 
= 25 -4V. 
From Eq. 1 the frequency is now given by 
f = 25-4 xlO 8 

~ 4x227x0-331xl5xl0 3 
The value of the feedback resistor Rf is given by 
R 2V CC -Vi n (2x28)-25-4 
Rf =-17~^ 94x10-3 =326a 
The nearest preferred value of 330O is used in the practical circuit. 

Page 287 


Transistor Spreads 

The drive transformer designed in the previous section is intended to 
drive a circuit containing low-a', high-Vbe transistors. 

The performance of the converter with transistors having high a' 
and low Vbe will be modified very slightly, except that the operating 
frequency might decrease by a maximum of 14% from the calculated 
value. The frequency may be adjusted, if required, by extracting 
a few laminations from the core of the drive transformer. With the 
components shown in the circuit in Fig. 1, and for typical transistors, 
the operating frequency is about 500c/s. 

For maximum spreads in transistor characteristics, the change in 
output voltage, output power and efficiency will be less than 3 %. 

Output Transformer 

If the magnetising current of the output transformer is to rise to 400mA 
during the half-cycle time t (equal to 1msec), the inductance required 
for each half of the primary, as given by Eq. 5, is 

1 x TO- 3 

L = 28 x A L * ™ = 70mH. 

400X10- 3 

To avoid excessive loss of power, the resistance of each primary 
winding should be less than 0-2Q. 

The peak collector current is the sum of the magnetising, feedback 
and load currents. Therefore it would appear that the circuit can 
operate up to 5-5A load current, provided the two halves of the circuit 
and the transistors are identical. In practice, because of slight unbalance 
in the circuit and the fact that the transistors are not matched, the 
out-of-balance current through the output transformer causes some 
premagnetisation of the core. As a result the collector current of one 
of the transistors will rise to a higher peak value than the other. If the 
circuit has been designed for operation up to the maximum ratings of 
the transistors, the peak collector current can thus exceed the safe 
value, if the circuit is not modified. 

The unbalance of the circuit can be reduced by using bifilar windings 
both for the primary of the output transformer and for the secondaries 
of the drive transformer. 

Spreads in a' are more difficult to deal with; the best method for 
obtaining balanced collector currents is to use a matched pair of 
transistors. It would then be possible to operate the circuit up to the 
full theoretical value of load current, 5-5A, with a consequent increase 
in output power of about 20 %. An external base resistor can be used 
to reduce the effect of spreads in Vbe. 

If no precautions are taken to avoid the unbalance, the load current 

Page 288 


must be limited to 4-5A; and the peak collector current must not 
exceed 5A, including the feedback current and the magnetising current 
of the output transformer. This allows for up to 1 A of out-of-balance 
current plus surges from the smoothing system. 

Even with these limitations, it is possible to obtain an output power 
of 100W with a 28V supply, with the additional advantage of using 
transistors with full spreads. 

With matched transistors and a purely resistive load, 130W output 
can be obtained at about 90 % efficiency. 

Starting Circuit 

On full load, the reflected load resistance is approximately 5 6CI. With 
a feedback turns ratio of 2, a minimum low-current a' of 20, and Rt>b 

IOQ 20O 300 400 500 600 
Load current (mA) 

Fig. 3 — Effect on operation of varying the load current 

equal to 35-6iQ, the minimum base current required for oscillation is 
(Eq. 6) given by 

I > 25X2 

b 20x5-6-2x35-6 
> l-23mA. 


The value of Rl, using a diode to initiate oscillation, is (from Eq. 7) 


Rl = 

= ll-3ka 

2x 1-23 x 10-3+0-02X10- 3 
The nearest lower preferred value of lOkQ should be used. 

Page 289 



With R2 equal to 3-30, a value of 3-3kO was found to be adequate 
for Rl for reliable starting. These values were found to be satisfactory 
for both choke input and purely resistive loads. Higher values for Rl 
might prevent starting with large capacitive loads, and lower values 
would reduce the efficiency. 


The performance of the converter shown in Fig. 1 is as follows 
Supply voltage 
Supply current 

Input power 


Ripple voltage 

Output voltage 

Output current 

Output power 


Over the range of temperature from 

10°C to +80°C the per- 

formance is hardly affected. Reducing the copper losses in the output 

Supply voltage (V) 

Fig. A — Effect on operation of varying the supply voltage 

transformer can lead to a higher output and an efficiency of about 

The effects of varying the load current and supply voltage over a 
wide range are shown graphically in Figs. 3 and 4. 


The heat sinks for the practical circuit (Fig. 1) can be made of com- 
mercial copper | in. thick of area 3f x 14 in. (conveniently folded), that 

Page 290 


is, about 3 -2111111 thick by 95 x 355mm. The surface should be blackened 
to assist cooling by radiation. An equivalent heat sink in aluminium 
could also be used. The thermal resistance of the heat sink, 0^ , should 
be about 2°C/W. The maximum ambient temperature at which the 
converter will operate satisfactorily is then approximately equal to 
80°C for a dissipation of 3W per transistor. 


Fig. 5 shows the collector voltage and current waveforms of the two- 
transformer converter working under full load conditions. The col- 
lector current waveform for a purely resistive load is shown in Fig. 5(b). 

(a) Collector voltage 

(-\ Collector current with 
0-5>jF across the load 

Collector current with 
(d) lOOuFcapacitance across 
the load 

Collector current witl 
(e) choke input filter as 
in Fig 4 

Fig. 5 — Collector voltage and current waveforms for various load conditions 

With a small capacitance across the load, the output transformer 
starts to ring. As a result the collector current (c) rises to a higher 
peak value. If the capacitance is much higher, the oscillation is damped 
and the collector current does not rise to such a high value (d). On no 
account must the peak-current rating of the transistors be exceeded. 
The disadvantage of a large capacitive load is that it can affect 
starting when it initially short circuits the load. However, a surge 
limiting resistor can be connected in series with the load, being pro- 
gressively short-circuited as the converter is switched on. 

Page 291 


So that the converter can operate satisfactorily with a large capaci- 
tance across the output, it is necessary either to reduce the load current 
or, much the better solution, to use a resistive or choke input filter. 
The collector-current waveform when the latter is used is shown in 
Fig. 7(e). The spikes at the beginning of the waveform are caused by 
the inductance of the transformer and choke, and must not exceed the 
peak-current rating. 


Satisfactory operation of the converter (Fig. 1), with resistive 
starting, can be obtained by reducing the value of Rl to 3-3k£l and 
replacing the diode Dl by a resistor, R2, of 3-3CL Performance figures 
for this modified circuit are given below 

Supply voltage . . 
Supply current . . 

Input power 
Output voltage 
Output current 
Output power 





193 V 




Voltage Doubler 

A voltage doubler is often required instead of a bridge rectifier. A 

Fig. 6 — Voltage-doubler output 

suitable circuit is given in Fig. 6. Results of measurements carried out 
on this circuit are displayed graphically in Fig. 7. 

Page 292 


The voltage doubler presents a large capacitive load, therefore a 
progressively short-circuited resistor is recommended to prevent large 
peaks of charging current from appearing immediately after switching 

SO lOO ISO 200 2SO 300 
Load current (mA| 

Fig. 7 — Effect of varying the load current on operation of vol tage-dou bier circuit 

on the supplies. This resistor can be in series either with the supply 
voltage or with the output-transformer connection and CI (see Fig. 6). 

Summary of Transformer Details 

Core material 


Primary winding 
Secondary winding 

H.C.R. alloy (Telegraph Construction and Main- 
tenance Co. Ltd.) pattern 224, 50 laminations 
Insulated Components and Materials Ltd. 187A 
227 turns of 34 s.w.g. enamelled copper wire 
57+57 turns (bifilar winding) of 30 s.w.g. enam- 
elled copper wire 


Primary windings Inductance = 70mH/winding 

Resistance < 0-2£J/winding (bifilar winding) 
Secondary winding Resistance < 15H 
Turns ratio 1+1:8-2 

Further Modification 

A modified circuit is shown in Fig. 8. Although the collectors can be 
connected to the same heat sink, or directly to chassis in equipment 
having the negative side earthed, the circuit operates as a push-pull 
common-emitter amplifier with the input applied between base and 
emitter. Fig. 8 is a redrawn version of Fig. 1, with a resistor R2 in 
place of Dl. The main difference between the two circuits is that 
the collector and emitter connections are interchanged; a separate 
starting circuit is used for each transistor. In this arrangement a diode 
must not be used in place of R2, because there would be no means of 

Page 293 


diverting base current and the transistor would never be cut off. 

The performance of this modified circuit is almost identical to that 
of the circuit in Fig. 1, with resistive starting, except that the efficiency 
is one or two per cent lower because of the additional current drain 

Fig. 8 — Further modification of circuit shown in Fig. 1. The emitter and collector 
connections are reversed and separate starting resistors provided for each transistor. 

arising from the separate biasing arrangements. The performance is 
as follows: 

Supply voltage 28V 

Supply current 

Input power 
Output voltage 
Output current 
Output power 









In push-pull d.c. converters, the peak voltage applied to the transistors 
in the cut-off state is twice the supply voltage. The supply voltage must 
be restricted to half the allowable peak collector voltage for the 

It is often desirable to operate the converters from a higher supply 
voltage than the simple push-pull circuit will allow. A bridge circuit of 
four transistors can then be used. 

A practical circuit with a suitable starting arrangement is shown in 
Fig. 9. 

Page 294 


The required drive is applied by a small saturating transformer Tl, 
in conjunction with a feedback resistor Rf . T2 has a linear character- 
istic and is used to step up the voltage to the value required for the 

Diagonally opposite transistors (Trl and Tr3, or Tr2 and Tr4) 
conduct together. Thus, when Trl and Tr3 are 'bottomed', the supply 
voltage will appear across Tr2 and Tr4, which are cut off. Therefore 
the voltage across any transistor will never exceed the supply voltage 
V C c , and this converter can be used with a supply voltage of twice the 
value allowed for any other push-pull arrangement. 

This converter, like the two-transistor versions, is suitable for variable 
loads. This is because the collector current in any transistor does not 
rise to the peak value determined by the drive but, as in the push-pull 

ri r-A/wv-H^l 

Tr4 Tr3 



" WW <$MSULr- 

•^pn pnf?^ 



Fig. 9 — Practical arrangement of bridge converter 

circuit already described, to a value equal to the load current plus the 
magnetising current of the output transformer and the feedback current. 

This converter, also, has good regulation. It is economical to con- 
struct, because it uses only a small saturating transformer with relatively 
inexpensive core material. The output transformer, being conventional, 
is relatively cheap for the output power obtained. 

OC28 transistors can be used in the suggested circuit with a 56V 
supply, and it is possible to obtain output powers of up to 200W, with 
an overall efficiency greater than 80%. The same transformer-design 
procedure can be adopted as for the other circuits. However, the 
circuit designer must still ensure that the ratings of the transistors are 
never exceeded. 


A single-transformer converter with capacitive-resistive timing is 

Page 295 


shown in Fig. 10. This should be particularly suitable for low powers. 
It operates on the same principle as the other circuits, the instant of 
switching being controlled by decreasing the base currents rather than 
by increasing the collector currents. 

A normal linear output transformer can be used. The circuit is 
complete as shown, except that the usual biasing arrangements must 
be added. Each transistor should have separate bias, and this can be 

Fig. 10 — Capacitive-resistive timing suitable for low-power converter 

provided by feeding the base from a potential divider connected across 
the supply. As the timing resistors R form the lower branches of the 
potential dividers, only two extra resistors are required, connected 
between the negative supply line and the base of each transistor. 

The drive is arranged such that the transistors operate in the bottomed 
or cut-off condition, so generating a square- wave output. 

In a circuit as shown in Fig. 10, two OC84 or OC83 could be expected 
to give an output of about 5W from a 12V supply. For even lower 
output powers a one-transistor circuit may be suitable, but here again 
switching will have to be controlled by the base current rather than 
by the collector current. 


The circuits described have distinct advantages over the one-transformer 
arrangements using a saturating transformer. Briefly these advantages 

(a) improved performance under varying load conditions with reduced 
stress on the transistors, even though operation up to the maximum 
ratings of the transistors is possible 

(b) transformer design is less critical, so permitting the use of one 
small saturating transformer and a larger, linear output trans- 
former. This is a cheaper solution than using one large saturating 

Page 296 



Where high-power transistors are used in non-mobile equipment, the 
supplies can most conveniently be taken from the mains. A mains- 
operated stabilised d.c. power unit is required, which is capable of 
providing low voltages at currents of the order of a few amps. This 
chapter is concerned only with the basic principles of stabilised d.c. 
power supplies, and more particularly with the provision of a protec- 
tion circuit. Although practical values are given, the final circuit will 
depend very much on particular requirements. 


The simplest voltage stabiliser is the emitter follower with the base 
connected to some reference potential, as in Fig. 1(a). In principle the 







" io 



1 < .. 










Fig. 1 — Simple form of stabiliser (a) and application of d.c. feedback (b) 

reference voltage may be taken from a standard cell or a Zener diode, 
or from a line stabilised by a (gas-discharge) voltage-stabiliser tube 
by means of a potential divider. The emitter follower divides the out- 
put impedance of the supply by approximately l+<x' . When a lower 
output impedance is required, for better regulation or to prevent feed- 
back through the supply, an amplifier can be inserted before the 
emitter follower and d.c. feedback applied, as shown in Fig. 1(b). 

The 'black boxes' of Fig. 1 may each represent one or more tran- 
sistors. Referring to the one-amp supply unit of Fig. 2, the amplifier 

Page 297 


o 6 

Page 298 


is the long-tailed pair (Tr3 and Tr4) and a compound emitter follower 
is used (Tr5 to Tr7), consisting of three transistors connected in cas- 
cade. The latter is no more than a device for effecting a greater reduc- 
tion in output impedance. 

The long-tailed-pair amplifier balances out the variations in base- 
emitter voltage with temperature. The effect of the variations in I co is 
made very small by using silicon transistors. The effects of temperature 
on the germanium transistors in the compound emitter follower are 
taken care of by the d.c. feedback loop, which extends from the emitter 
circuits of Tr5, Tr6 and Tr7 to the base of Tr4 and thence from the 
collector of Tr4 to the base of Tr5. 

The reference voltage is applied to the base of Tr3 and is derived 
from the stabilised —85V fine by a potential-divider chain. RV3 is 
ganged to the variable input transformer which supplies the emitter 
follower, to limit the power dissipation in the output transistor. A 
Zenith V544 is suitable. The coupled emitters of the long-tailed pair 
are run from a stabilised positive fine rather than from the earth 
line, so that control can be maintained right down to zero output 
voltage. The positive voltage is stabilised by a second 85A2. For high 
gain, and to render the output voltage independent of the input, the 
lOOkQ collector load is connected to the —85V stabilised line. 

The lkQ resistor in series with the 0-1 (xF capacitor shunts the 
output of the long-tailed pair, and reduces the gain of the amplifier 
at the frequencies ( ~ lOkc/s) at which the phase shift causes 

The series transistor Tr7 operates at a high dissipation and requires 
mounting on a substantial heat sink. 


High-power transistors often need to be operated near their peak 
current and voltage ratings, and so can be destroyed by a sudden 
overload, such as may occur in an experimental circuit under fault 

The overload protection circuit about to be described can switch 
off the supplies in less than 50(xsec. The transistors in the stabilised 
power supply itself are protected against overload and short-circuit 
conditions. In addition, the current at which the overload protection 
operates can be set well below the available output from the power 
supply, and can be varied over a wide range. Thus experimental cir- 
cuits, powered by the stabilised supply, can also be protected against 
their own shortcomings. This feature is especially useful for laboratory 
power supplies where experimental circuits may need to be protected 
against such conditions as thermal runaway. 

Page 299 



Page 300 


Fig. 3 shows how such a protection circuit (Trl and Tr2) may be 
incorporated in a power supply of the type shown in Fig. 2. In fact 
Fig. 3 differs from the previous one in a number of other respects : thus 
the stabilised lines are ±75V; and in the compound emitter follower, 
the final step in the cascade consists of three OC28 in parallel, so that 
the unit can supply output currents of up to 3A. 

The protection circuit is an Eccles- Jordan bistable circuit. Switch 
off is effected by cutting off the emitter follower by means of a small 
positive voltage applied to the base of Tr5. Tr2 is normally in the 
cut-off condition. The OA202 diode, which being a silicon diode has 
a low leakage current, ensures that changes in the leakage current of 
Tr2 do not affect the stability of the supply. In normal operation the 
anode of the diode is more negative than the maximum negative 
voltage on the base of Tr5. 

In the emitter circuit of Trl is a low resistance (Rl) through which 
the load current flows, and across which is developed a voltage propor- 
tional to the load current. The circuit is arranged so that, when the 
load current exceeds a certain value, the bistable circuit is triggered 
into its other state. Tr2 bottoms, causing the OA202 to conduct. The 
base of Tr5 is now at a small positive voltage with respect to the zero- 
voltage line, and the emitter follower is therefore cut off. 

The emitter of Tr2 is connected to a positive voltage (+2V) to allow 
for the small voltage drop across Tr2 and the OA202. This arrangement 
ensures that the emitter follower can be definitely cut off. The base of 
Tr2 is connected to a voltage (+10V) more positive than that at its 
emitter (+2V), to maintain Tr2 in the cut-off state, when not triggered, 
over the whole range of ambient temperature (20°C to 45°C) for which 
the circuit is designed. 

The current through the potential divider consisting of the 47Q, 
220Q and 2-2kQ resistors must be relatively large (40mA) because, 
when the circuit triggers to its other stable state, the current from the 
base of Tr5 will flow through the bottomed transistor Tr2 and through 
the 47Q resistor to the common line. If the latter current is comparable 
with that through the potential divider, it can inhibit the change-over 
action of the bistable circuit after that circuit has been triggered. 

After a fault has been cleared, C r can be connected momentarily to 
the collector of Trl by a spring-loaded switch, to reset the circuit. 
The charging current of C r produces a positive triggering voltage at 
the base of Tr2, cutting off this transistor. The switch is shown in the 
reset position in the circuit diagram. 

Practical Details 

The complete circuit (Fig. 3) is given by way of example only; the 
effects of component tolerances have not been considered. The power 

Page 301 


supply is designed to provide an output current of up to 3A at a 
stabilised voltage of 27V. The protection circuit operates at the 3A 
maximum output. 

The protection circuit needs a triggering voltage, developed across 
Rl, of approximately IV. Different values of Rl can be switched in to 
vary the load current at which triggering occurs. For an output of 1A, 
Rl should be 10; for 2A, Rl should be 0-5O; and so on. Approximate 
values can be used for Rl, and RV2 adjusted so that triggering occurs 
at the exact value of required load current. 

If there is a fault in the load, the discharge current of the output 
capacitor (limited only by the resistance of the fault) is added to the 
current flowing through the output terminals. The combined current 
pulse could damage the circuit in which the fault has occurred. For this 
reason, the output capacitor may need to be reduced to 100>F. 

The protection circuit is inoperative until C r has been charged. The 
value of C r is dependent on the value of the output capacitor, and 
4f/.F is the minimum for a 1000>F output capacitor. C r may be reduced 
if a lower value is used for the output capacitor. 

If the stabiliser circuit is triggered 'on' with a fault still present, the 
protection circuit is inoperative for a few milliseconds; and the dissi- 
pation in the series transistors could become excessive, because the 
full unstabilised voltage acts across the output transistors, causing a 
large current to flow through them. A double-pole switch is therefore 
used for the reset operation, so that one output terminal is disconnected 
at the same time as C r is connected across Trl. 

Tr6, Tr7, Tr8 and Tr9 need to be adequately mounted to ensure 
that they are within the junction-temperature rating. 

Expected Performance 

Basically, changing the value of Rl adjusts the triggering current 
between 20mA and 3A. With Rl equal to 10, the variable resistor can 
be adjusted, for example, until the circuit triggers at exactly 1A. The 
variation of a given triggering current over a temperature range of 
20 to 45°C should be less than 10%. The value of triggering current 
should be nearly constant for all settings of output voltage down to 
less than IV. 

If the supply is short circuited when set to an output voltage of 25 V, 
and the circuit set to trigger at 1A, the peak current is expected to be 
2A and the time of the pulse about 25[j.sec. 

Page 302 



A number of high-frequency measurements can be made using an r.f. 
input signal modulated at a low frequency (lOOOc/s). The r.f. signals 
at various points in the circuit can be measured by demodulating 
with a diode detector, and feeding the resultant low-frequency signal 
into a high-gain tuned amplifier. The output from the amplifier is 
rectified and the d.c. output fed to a moving-coil indicator meter. This 

RS _, 2V 



T1 wound on Vinkor former LA2303. Main winding 667 turns tapped at 230 turns from supply 

end, feedback winding 1 5 turns, output winding 1 36 turns centre-tapped, all of 38 s.w.g. enamelled 

copper wire. T2 wound on Vinkor former LA2103. Primary 104 turns centre-tapped, secondary 

538 turns tapped at 86 turns from chassis, both of 34 s.w.g. enamelled copper wire. 

Fig. 1— 1000c/s oscillator 

method obviates the use of a high-gain r.f. amplifier, which would 
have to be either wideband or tunable to the r.f. signal to be detected. 


Most signal generators either have a high input impedance ( ~ 1MQ) 
requiring an input of 30 to 50V, or a low input impedance ( ~ 600Q), 
requiring a relatively high input power ( - 100m W). In the circuit 
shown in Fig. 1, these requirements are met by outputs (2) and (1). 

Page 303 


vO C .~ 
J; 3 O 

.E.S « 

;; rt ° 
£ £ « 



« fl» c 

— i. bO 



<£ x,o 

2 E" 

O Ug 


3 fe *• 
m " 


^ ■»! 
o F"*" 



_J 3 « Q. 
. <JT> E 

UX 11 19 
> Q.-0 

ON i u 

* 5 *& 

.. o F-Sl 

IX 3 ? O 

J>T3 OO 

O * fc « 

a E„-s? 


* ■ -- m 



Page 304 


A low-power sinusoidal oscillator drives a class B push-pull amplifier. 
Class B operation is preferred in the interests of low battery drain. 
Negative feedback is applied to the push-pull output stage by R4, 
which restricts the mutual conductance of each transistor, during its 
conduction period, to 45mA/V. 

The oscillator provides a voltage feed to the output transistors 
which, having a fixed mutual conductance, deliver an output current 
independent of the transistor input impedance. The input impedance, 
however, can vary over a very wide range, and to avoid the possibility 
of the oscillator squegging under certain conditions, the oscillator 
tuned circuit is damped by R3. 

The maximum power dissipation in the output transistors occurs 
if the output is short-circuited. The transistors then pass peak currents 
of 110mA, resulting in a dissipation of 300m W per transistor if the 
maximum supply voltage is 12V. Using OC83 transistors mounted 
on a heat sink of 7 x7cm of 16 s.w.g. aluminium per transistor, this 
dissipation allows a maximum ambient temperature of 55°C. The 
output transistors should be bolted down to their heat sinks by means 
of standard 20mm cooling clips (obtainable from Kimber-Allen Ltd. 
or distributors). 

C3 tunes the secondary of the output transformer, thus reducing 
the higher-harmonic content, and permitting a lower inductance and 
hence lower losses. 

The output stage can deliver a maximum output power of 120mW 
(into a resistance of 600Q) at a nominal supply voltage of 10V. The 
total-harmonic distortion under all conditions is less than 1 %. The 
output voltage will decrease slightly as the load increases, because of 
the effect of R4. 

At a supply voltage of 10V, the remaining performance figures are 
as follows. The direct-current consumption is 14mA with no load, 
30mA on full load, and 51mA with the output short-circuited. The 
frequency variation with temperature is less than 200 p.p.m. per °C. 
The frequency variation with supply voltage is 250 p.p.m. per volt. 
The output voltage from output (1) is 9-7V r.m.s. with no load, and 
8-7V r.m.s. on full load. The output voltage from output (2) is 6-25 
times that from output (1). 

All resistors should be ±10%. Rl, R2, R3 and R4 should be JW, 
and R5 should be |W. CI should be ±20%, C2 and C3 should be 
±1 % with polystyrene dielectric, and C4 should be 12V d.c. wkg. 


The lOOOc/s tuned amplifier is shown in Fig. 2. To obtain a good 
signal-noise ratio, the first stage has an emitter current of 200(xA. All 

Page 305 


the other stages have emitter currents of about 1mA. The primaries of 
T2 and T4 are tuned to lOOOc/s to reduce the bandwidth of the amplifier 
and thus reduce noise. The working Q of the tuned circuit of T2 is 
approximately 100. The working Q of the tuned circuit of T4 is about 
10, but depends to some extent on the signal level supplied to the 
detector. The turns ratio of T4 is chosen so that, when the transistor 
overloads at full supply voltage, the meter current is about 2mA ; thus 
the meter cannot be excessively overloaded. The time-constant of the 
detector circuit is normally 100 milliseconds, but can be increased, 
when working at very low levels, to one second, by switching in the 
1000(xF capacitor. 

RV15 is a variable attenuator covering to 20dB. Switched attenu- 
ation is provided in three sections. The first two 20dB steps of switched 
attenuation are placed between the second and third stages. In order to 
prevent overloading of the preceding stage, the second two steps of 
20dB attenuation are placed between the first and second stages. The 
final two steps precede the first stage. The distribution of the attenua- 
tion in this way, rather than by putting it all at the input to the 
amplifier, ensures that the input voltages to the first and second stages 
are always the maximum possible, resulting in the best possible signal- 
noise ratio. Each stage is individually decoupled, to prevent oscillation 
and the feedback of signals through the supply. The stability margin 
round each possible feedback loop is about 20dB at lOOOc/s. 

Typical performance for OC71 transistors is as follows: The maxi- 
mum power gain is 150dB, assuming the detector circuit to be replaced 
by a resistor. The maximum current gain — the direct output current 
divided by the r.m.s. input current — is 2 x 10 6 . The equivalent noise 
input current is 0-2mfjiA. The bandwidth is lOc/s. The centre frequency 
drift with supply voltage is less than +500 p.p.m. per V. With tempera- 
ture, the main source of drift is the tuned circuit coil which will cause 
a frequency shift of less than —300 p.p.m. per °C. The centre-frequency 
accuracy is ± 1 %. Power consumption is approximately 4mA at full 
supply voltage (16 -5V). The estimated life of an Ever Ready 16 -5V 
grid-bias battery is several hundred hours when used for 4 hours per 
day, after which the voltage will have dropped to 10V. 

Page 306 


Alloy-diffused transistor 
Alloy-junction transistor 
Avalanche effect 




Barrier potential 115 

Base resistance rbt>' . . 76 

Bass response . . 126 

Bi-directional operation . . 4 

Bottoming . . 64, 124, 243, 250 

Carrier (hole) storage . . 107, 158, 258 
Catching diode . . 254, 261 

Coil loss 188 

Collector capacitance . . 79 

„ junction . . 3, 245 

„ resistance 79 

Common electrodes . . 33, 53 

Coupling capacitance . . 126 

Crystal-controlled oscillator . . 213 
Current ampl. factor 16-17, 35, 38, 40, 

48, 75, 90 

gain . . . . 35, 38, 126 

„ saturation . . . . 243 

Cut-off frequency 41, 89, 120, 193-4 

Dark current 
Delayed switch-off 
Depletion capacitance 

„ layer . . 
Detection efficiency 
Diffusion capacitance 
Distortion . . 99, 

„ crossover . . 
„ input-circuit 
„ non-linearity 

Early effect 

Emitter capacitance c e . . 



-2, 253 


. . 74, 76 



61, 94, 124 

102, 123, 127, 

156-8, 228 


.. 15, 157 

.. 16, 156 



Emitter decoupling 
„ efficiency 
„ follower 
,, junction 
,, resistance r e 

Factor of stability 
Feedback characteristic 

„ factor (a 
Forward characteristic 

„ recovery 
Frequency response 

„ stability 

Grounded electrodes . 

Half-supply- voltage principle. . 62, 136 

Heat sink 95-99 

High-frequency parameter f i . . 91 

Hole 2 

„ (carrier) storage . . 107, 158, 258 

„ -storage time-constant 215, 252 

h parameters . . . . . . 23 


40, 297 



56, 57 

33, 53 

Input characteristic 


„ impedance 

34,37, 128 

,, resistance. . 

. . 14, 196 

Insertion loss 




Junction temperature . . . . 67, 94 

Knee 18, 62, 105, 123, 152-4 

Large-signal equiv. circuit . . 245 
Leakage current 19-20, 50, 67, 105 
Light current .. .. 115 
„ sensitivity. . . . 8, 115 
Loop gain 221,231 

Page 307 

Majority carriers 2 

Matched transistors 156, 169, 173, 288 
Minority carriers . . . . 2 

Modified z parameters . . . . 26 

Mounting-base temperature . . 97 

„ position . . . . 8 

Mutual conductance . . 83, 194, 231 

Negative feedback 


N-p-n transistor 

N-type material 

Output characteristic 
„ impedance 
„ resistance 

P-n-p transistor 
Polarity . . 
Power gain 
P-type material 

RC coupling 
Recovery time 
Rectification efficiency 
Relay switching . . 
Reverse characteristic 




. . 34, 38 


43, 90, 235 
.. 8,115 

. . 7, 103 
. . 36, 39 


. . 29, 42 

2,74, 115,251 





Schmitt triggering 
Storage capacitance 
„ temperature 
Supply decoupling 
Symmetrical transistor 

Temperature effects 
„ stability 

Thermal resistance 
„ runaway 
„ stability 

T network 

Transfer characteristic 

Transformer coupling 




. . 52, 123 

229, 233 









62, 125, 149 

. . 29, 82 


.. 28,42 

Turn-on transient forward voltage 108 

Unilateralisation 181, 184 

Voltage ampl. factor . . . . 36, 39 
„ gain 35, 39 

y parameters 30 

Zone refining 4 

z parameters . . . . . . 26 

Page 308